cs.CL [Total: 24]
cs.CV [Total: 66]
cs.HC [Total: 1]
cs.LG [Total: 6]
cs.GR [Total: 1]
cs.CR [Total: 1]
cs.AI [Total: 11]
cs.SE [Total: 1]
cs.RO [Total: 4]

cs.CL [Back]

[1] Mean-Field Dynamics of Chain-of-Thought Reasoning in Large Language Models cs.CL | cs.AIPDF

Hao Ai

TL;DR: 本文提出一个理论框架，用于解释大语言模型（LLM）中思维链推理的统计规律，而不简化模型架构或类比物理系统。它将推理建模为线索图上的引导发现过程，并通过平均场近似推导出描述已发现线索比例的一维常微分方程。实验通过学生LLM对教师LLM输出的归一化惊异值识别线索标记，并在多个思维链上平均得到统计规律，结果表明这些规律可复现且能用所提理论方程拟合。

Details

Motivation: 旨在为LLM的思维链推理行为提供理论解释，以加深理解并指导模型优化，避免对模型架构进行简化或依赖现有物理系统类比。

Result: 实验表明，在同一数据集内获得的统计规律具有可复现性，并且可以通过求解提出的理论方程进行拟合。

Insight: 创新点在于将LLM推理形式化为线索图上的引导发现过程，并应用平均场近似推导出可解释的微分方程模型，为理解思维链的动态提供了新的理论视角。

Abstract: Large language models (LLMs) with chain-of-thought reasoning have been widely applied in recent years, and theoretical explanations of their behavior may help deepen our understanding and guide model optimization. In this study, we introduce a framework that seeks statistical regularities and theoretical interpretations in LLM reasoning without simplifying the model architecture or making analogies to existing physical systems. We formulate LLM reasoning as a guided discovery process on a clue graph, and derive a one-dimensional ordinary differential equation for the fraction of discovered clues using the mean-field approximation. Experimentally, clue tokens are identified using the normalized surprisal of a student LLM on the outputs of a teacher LLM, and statistical regularities are obtained by averaging over many reasoning chains of thought. Our experiments show that the resulting statistical regularities are reproducible within the same dataset and can be fitted by the solving the proposed theoretical equation.

[2] Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support cs.CL | cs.AIPDF

Gary Simethy, Daniel Ortiz Arroyo, Petar Durdevic

TL;DR: 本文研究了三种将大型语言模型（Qwen2.5-32B-Instruct）与可解释的废水处理模拟器（CCSS-IX）相结合的方法，以支持工业因果推理。这三种方法分别是：实时模拟器调用、结构化参数注入和解耦式召回-推理检索器。在废水处理因果问答基准测试中，三种方法均显著优于检索增强基线，其中解耦式召回-推理检索器在跨工厂迁移和反事实推理方面表现出色，并在通用知识问答任务上验证了其泛化能力。

Details

Motivation: 废水处理操作员需要基于其工厂变量间的具体相互作用和动态传播来回答因果问题，而非依赖通用预训练文本。现有方法难以满足这种对特定、动态工业知识的精确推理需求。

Result: 在198个因果问题基准上，三种方法准确率分别为99.5%、79%和75.8%，远超最强检索增强基线（48%）。解耦式召回-推理检索器在跨工厂迁移后仍达88%准确率，在60个反事实问题上比结构化参数注入方法提升16.3个百分点。在AI2推理挑战赛上，该检索机制达到79%准确率，优于未约束的Llama-3.1-8B（76%）和全注入方法（74%）。

Insight: 论文创新性地系统比较了实时工具调用、静态参数注入和学习型数值参数检索三种工业知识落地范式，并提出了轻量、可迁移的解耦式召回-推理检索器架构。其核心洞察在于，将知识检索与推理解耦，并学习从文本查询到模拟器数值参数的映射，能有效支持动态因果和反事实推理，且该机制具有跨领域泛化潜力。

Abstract: Wastewater operators need answers grounded in how their plant’s variables interact and how fast effects propagate, not in generic pretraining text, when asking causal questions such as “why is N2O rising?” or “what happens if I cut aeration by 20%?”. We compare three concrete ways to ground a frozen Qwen2.5-32B-Instruct model in an architecturally interpretable wastewater simulator (CCSS-IX): a live simulator oracle (Method 1), structured parameter injection (Method 2), and a Decoupled Recall-Reasoning (DRR) retriever (Method 3). On a 198-question causal benchmark the three reach 99.5%, 79%, and 75.8%, forming a deployment ladder above the strongest retrieval-augmented baseline at 48%. The DRR retriever has 110M parameters and trains per plant in ~17 seconds; after cross-plant transfer to a biologically distinct plant it still reaches 88%, while Method 2’s static table cannot transfer. On a 60-question counterfactual benchmark only Method 3 handles queries about what happens after an intervention: +16.3 pp over Method 2, paired 95% CI [+7.1, +26.4] pp, with 100% on the timescale and operating-regime categories. On the AI2 Reasoning Challenge (ARC) with an OpenBookQA fact corpus, the same selective-retrieval mechanism reaches 79% versus unconstrained Llama-3.1-8B 76% and full-injection 74%, a +3 pp out-of-domain replication that argues against a result specific to wastewater treatment. We provide the first single-simulator comparison of live tool-use, static parameter injection, and learned numerical-parameter retrieval for industrial causal question answering.

[3] RIG-RoPE: Relation- and Instance-Gated Rotary Positional Encoding with Duration-Aware Temporal Coordinates cs.CL | cs.CVPDF

Donggen Li

TL;DR: 本文提出了RIG-RoPE，一种针对交错多模态上下文（如文本、图像、视频）改进的旋转位置编码机制。它通过引入模态指示符、视觉实例标识符和信息持续时间坐标，解决了静态多维位置分配中的两个关键问题：跨模态/实例的空间干扰，以及不同模态间信息密度差异导致的时间相位处理不当。该方法无需学习参数，可直接在分块注意力内核中实现。

Details

Motivation: 动机在于解决现代多模态大语言模型中，静态多维旋转位置编码（如M-RoPE）在交错多模态上下文中的局限性。具体问题包括：1）高度/宽度旋转可能被错误地应用于空间位移未明确定义的跨模态或跨实例的token对，导致空间干扰；2）时间坐标通常被当作等步长计数器处理，忽略了文本、图像、视频token在信息密度上的巨大差异。

Result: 本报告为初步研究，建立了方法的公式化和验证路径，并未声称经验上的优越性。因此，摘要中未提及具体的定量实验结果、基准测试或与SOTA模型的比较。

Insight: 创新点包括：1）提出关系与实例门控机制，仅对同一视觉实例内的查询-键对应用高度/宽度旋转，否则将未知空间位移边缘化而非设为零；2）引入持续时间感知的时间坐标，根据文本（单位持续时间）、图像（基于维度的对数空间尺度）和视频（在有效帧上的对数时间扩展）的信息密度，使用插值的累积块持续时间进行时间旋转。此外，还提供了关于避免普通跨实例空间旋转的规范不变性论证、共享高度/宽度子空间下静态ID不可能性结果，以及反对等步长多模态时间的持续时间一致性论证。

Abstract: Rotary positional encoding (RoPE) is a core component of modern language models and has been extended to multimodal LLMs through multidimensional variants such as multimodal RoPE (M-RoPE), which split positional channels into temporal, height, and width subspaces. This report identifies two limitations of static multidimensional position assignment in interleaved multimodal contexts. First, height/width rotations may be applied to token pairs whose spatial displacement is not a well-defined geometric object, producing cross-modal and inter-instance spatial interference. Second, temporal coordinates are often treated as equal-step counters, so a text token, an image block, and a video segment can advance the temporal phase by comparable amounts despite different information density. We propose RIG-RoPE, a relation- and instance-gated RoPE mechanism with duration-aware temporal coordinates. RIG-RoPE augments each token with a modality indicator, a visual instance identifier, and a scalar information-duration coordinate. It enables H/W rotations only for query-key pairs from the same visual instance; otherwise the unknown spatial displacement is marginalized rather than set to zero. Temporal rotations use interpolated cumulative block durations: text tokens consume unit duration, images use a dimension-aware logarithmic spatial scale, and videos further apply a logarithmic temporal extension over effective frames. We provide a gauge-invariance argument for avoiding ordinary cross-instance spatial rotation, an impossibility result for static IDs under shared H/W subspaces, and a duration-consistency argument against equal-step multimodal time. RIG-RoPE adds no learned parameters and can be implemented inside tiled attention kernels with constant additional metadata per token. This preliminary report establishes the formulation and validation path without claiming empirical superiority.

[4] Cross-Architecture Steering Transfer in Language Models: A Systematic Empirical Study cs.CL | cs.LGPDF

Ayushi Agarwal

TL;DR: 本文首次系统评估了跨模型引导转移，研究表明在满足规模条件时，独立训练的大语言模型间共享的几何表示具有功能性利用价值。通过分析五个不同架构和参数规模（0.8B-8B）的模型在15个语义域上的特征对齐，发现参数规模约1.7B是一个关键阈值，超过该阈值时跨模型特征对验证成功率达47-49%，且跨模型引导向量在监督概念上的胜率可达71.0%。

Details

Motivation: 探究独立训练的大语言模型之间共享的内部语义表示几何相似性是否具有功能性后果，即能否实现跨模型的行为控制，而无需微调。

Result: 在参数规模≥1.7B时，跨模型特征对验证成功率为47-49%（Pearson r ≥ 0.60，Procrustes余弦相似度0.895-0.956）；跨模型引导向量（B3-TI）在15个监督概念上的胜率达71.0%，优于同模型原生向量的68.0%；单个通用向量在5个模型中的4个上达到67.3%胜率，无需针对每个模型的监督。

Insight: 研究揭示了参数规模阈值（约1.7B）对跨模型引导转移的重要性，为机制可解释性工具在不同规模模型间的可迁移性提供了实证依据；同时，为“柏拉图式表示假说”提供了首个功能性补充，表明在特定规模条件下，独立训练LLMs的几何收敛支持无需微调的跨模型行为控制。

Abstract: Independently trained large language models may develop shared internal representations of semantic concepts despite architectural differences – but whether this geometric similarity has functional consequences for cross-model behavioural control remains untested. We present the first systematic evaluation of cross-model steering transfer and show that shared LLM geometry is functionally exploitable, conditionally: concept directions from one model can steer a different independently trained model when sufficient representational capacity exists. We study five open-weight models spanning three parameter scales (0.8B–8B) and two architectural lineages, training one Sparse Autoencoder per model across 15 semantic domains and testing alignment across all 20 directed model pairs. We observe a suggestive discontinuity near 1.7B parameters: at >= 1.7B scale, 47–49% of cross-model feature pairs validate (Pearson r >= 0.60, Procrustes cosines 0.895–0.956), while alignment degrades sharply below 0.8B. Cross-model steering vectors (B3-TI) achieve a 71.0% win rate across 15 supervised concepts versus 68.0% for same-model native vectors; a single universal vector achieves 67.3% in 4 of 5 models without any per-model supervision. Transfer degrades for models below 1.7B and for one model with generation instability, confirming that functional exploitability requires sufficient representational capacity. Our findings underscore the importance of scale thresholds in mechanistic interpretability: tools validated at 7B scale may not transfer to smaller models without revalidation. We provide the first functional complement to the Platonic Representation Hypothesis – geometric convergence across independently trained LLMs supports cross-model behavioural control without fine-tuning, under the identified scale conditions.

[5] Conditional Cognitive Biases in LLMs: How Biased User Turns Modulate In-Context Reasoning cs.CL | cs.CY | cs.HC | cs.LGPDF

Sachini Weerasekara, Sagar Kamarthi, Jacqueline Isaacs

TL;DR: 本文评估了最先进的指令调优大语言模型在多轮交互场景下认知偏见的表达情况。研究提出了一个新颖的三条件实验框架，用于区分用户偏见轮次暴露效应与语义内容效应，并构建了一个包含24,300个经过陪审团验证的用户提示的基准数据集，覆盖了9x9目标-人类偏见交互矩阵的所有81个单元。在八个前沿LLM上的实验表明，有偏见的对话上下文会系统性地增加其中六个模型的偏见表达。研究揭示了两种竞争性行为动态：接触偏见推理通常会放大下游偏见倾向，而明确的偏见线索则常触发对齐相关的抑制行为，从而减少显性偏见表达。

Details

Motivation: 旨在评估LLMs在现实多轮交互中认知偏见的表达，并厘清用户偏见输入（暴露效应）与其语义内容对模型后续推理的不同影响。

Result: 在八个前沿LLM上，相对于零样本基线，有偏见的对话上下文在六个模型中系统性地增加了偏见表达。研究构建了一个包含24,300个提示的基准数据集，并识别了偏见放大与对齐抑制两种竞争动态。

Insight: 创新点在于提出了一个能分离暴露效应与语义内容效应的三条件实验框架，并构建了大规模、结构化的偏见交互基准。客观来看，其对“偏见暴露”与“偏见内容”的分离设计，以及对模型行为中“放大”与“抑制”两种竞争动态的识别，为理解和缓解LLM的上下文条件偏见提供了新的方法论和深刻见解。

Abstract: We present an evaluation of cognitive bias expression in state-of-the-art instruction-tuned LLMs under realistic multi-turn interaction settings. Our work introduces a novel three-condition experimental framework that disentangles the effect of exposure to a biased user turn from the effect of the turn’s semantic content, alongside a benchmark of 24,300 jury-validated user prompts spanning all 81 cells of a 9x9 target-human bias interaction matrix. Across eight frontier LLMs, we find that biased conversational context systematically increases bias expression relative to zero-shot baselines in 6 of 8 models. We identify two competing behavioral dynamics underlying this effect: conversational exposure to biased reasoning generally amplifies downstream bias tendencies, while explicitly stated bias cues often trigger alignment-related suppression behaviors that reduce overt bias expression. We release our framework, codebase, and dataset to support future research on context-conditioned cognitive biases and behavioral adaptation in LLMs.

[6] ConWriter: Transition-Constrained Stateful Long-Form Story Generation with Lightweight Neuro-Symbolic Consistency Control cs.CLPDF

Jindong Li, Yang Yang, Zihao Liu, Yutao Yue, Menglin Yang

TL;DR: 本文提出了ConWriter，一个无需训练的框架，用于解决长故事生成中的叙事一致性问题。该方法通过场景级增量写作，结合静态故事要求、动态叙事记忆、符号状态推理和不确定性感知风险信号，在生成过程中进行一致性控制，防止局部错误传播。

Details

Motivation: 现有基于提示的长故事生成方法在故事增长时容易累积时间、事实、角色、常识和风格上的错误，缺乏有效的跨长上下文叙事一致性保持机制。

Result: 在ConStory-Bench基准测试的四个长故事任务（续写、生成、扩展、完成）上，针对Qwen3.5-Plus、DeepSeek-V4-Flash和GPT-5系列模型，在3k、6k和12k目标长度下进行了评估，实验遵循官方协议。

Insight: 将长故事生成从单一自由解码过程重构为基于状态演化和过渡约束的增量过程，并引入轻量级神经符号一致性控制，在生成早期进行验证和局部修复，是一种新颖的框架设计思路。

Abstract: Long-form story generation requires models to preserve narrative consistency across extended contexts, yet existing prompting-based methods often accumulate temporal, factual, character, commonsense, and stylistic errors as the story grows. We propose ConWriter, a training-free framework for consistency-aware long story generation. ConWriter writes stories incrementally at the scene level, guided by static story requirements, dynamic narrative memory, symbolic state reasoning, and uncertainty-aware risk signals. Rather than treating long-story generation as a single free-form decoding process, ConWriter maintains evolving story states, checks whether new scenes satisfy required narrative transitions, and uses uncertainty-aware risk signals to prioritize validation and localized repair. This enables consistency control during generation, before local errors propagate into later scenes. We evaluate ConWriter on ConStory-Bench, covering four long-story tasks: continuation, generation, expansion, and completion. Due to the high cost of long-form generation and evaluation, we use the first five cases from each task and test 3k, 6k, and 12k target lengths across Qwen3.5-Plus, DeepSeek-V4-Flash, and GPT-5 series. Experiments follow the official ConStory-Bench evaluation protocol.

[7] Position: It’s Time to Optimize LLMs for Self-Consistency cs.CL | cs.AIPDF

Itamar Pres, Belinda Z. Li, Laura Ruis, Zifan Carl Guo, Keya Hu

TL;DR: 这篇立场论文提出将自洽性作为理解和改进语言模型（LM）行为失败（如奉承、逻辑泛化不完整、自信但错误）的框架。作者认为，当前建模假设（即行为可在单个输出对上独立指定和评估）是这些失败的根源，并主张通过优化自洽性来统一解决多种问题。

Details

Motivation: 动机在于解决语言模型中持续存在的关键失败，如对用户框架的过度条件化（奉承）、不完整的逻辑泛化以及自信但错误的响应，这些失败难以通过仅评估单个输出对来检测。

Result: 论文未提供具体实验结果，但指出许多针对特定方面（如对抗鲁棒性和事实一致性）的改进技术可视为“一致性优化”过程的特例，并可用标准优化工具处理。

Insight: 创新点在于将自洽性作为统一框架来诊断和优化语言模型行为，提出通过一致性优化实现新模型属性，并探讨开发一般自洽语言模型的潜力和挑战。

Abstract: Despite ever-increasing sophistication in language model (LM) pre- and post-training pipelines, many important failures persist: models overcondition on user framing (“sycophancy”), exhibit incomplete logical generalization, and produce confident but incorrect responses. We argue that these failures arise from a modeling assumption permeating all aspects of the pipeline: that behavior can be specified and evaluated independently on single-output pairs. Many model failures are difficult, if not impossible, to detect without reasoning about relationships between a model’s responses across inputs. In this position paper, we propose self-consistency as a framework for understanding these failures. We first observe that a wide variety of techniques designed to improve specific aspects of LM behavior-targeting properties as diverse as adversarial robustness and factual coherence-can be understood as special cases of a common “consistency optimization” procedure and addressed with a standard set of optimization tools. We next outline a set of new model properties that could be achieved by optimizing for consistency, and conclude with a discussion of what it would mean to develop generally consistent LMs, including the capabilities they would enable and the objections they raise.

[8] DREAM: LLM-based Dynamic Role-playing via Event-Aware Memory Graph cs.CL | cs.AIPDF

Zhihao Xiao, Mengting Li, Xintao Wang, Linfeng Li, Limin Shui

TL;DR: 论文提出了DREAM框架，一种基于事件感知记忆图（EMG）的结构化记忆框架，用于增强角色扮演代理（RPA）的长期叙事和人格一致性。该框架将非结构化文本转换为时序有序、因果关联的事件图，构建动态双粒度角色档案，以捕捉稳定人格特征和事件驱动的行为演变。

Details

Motivation: 现有角色扮演代理主要依赖静态角色描述和非结构化记忆，难以维持长期叙事和人格一致性。论文旨在解决角色扮演中需要时序一致和因果基础的行为推理问题。

Result: 在CoSER、LIFECHOICE和提出的TCM基准测试中，DREAM实现了最先进的性能，优于多个强基线模型。

Insight: 创新点在于借鉴ABC认知模型，将非结构化文本结构化表示为事件感知记忆图，从而支持动态角色档案构建。这增强了角色扮演代理的可解释性和一致性，为基于LLM的叙事生成提供了新思路。

Abstract: Role-playing agents (RPAs) have emerged as a key application of large language models, enabling immersive and high-fidelity character simulation. Accurate role-playing of established characters requires not only stylistic imitation but also temporally consistent and causally grounded behavioral reasoning. However, existing RPAs primarily rely on static character descriptions and unstructured memory, limiting their ability to maintain long-term narrative and personality coherence. We introduce DREAM, a structured memory framework for role-playing agents inspired by the Activating Event-Belief-Consequence (ABC) cognitive model. DREAM transforms unstructured literary text into an Event-aware Memory Graph (EMG) that organizes character experiences into temporally ordered and causally linked event graph. This representation enables the construction of dynamic, dual-granularity character profiles that capture both stable personality traits and event-driven behavioral evolution. We further propose the Temporal Causal Memory (TCM) benchmark to evaluate temporal consistency and long-range causal narrative coherence. DREAM achieves state-of-the-art performance across CoSER, LIFECHOICE, and TCM, outperforming multiple strong baselines. Our approach demonstrates the effectiveness of structured memory in enhancing the interpretability and consistency of role-playing agents.

[9] Constraint-First Reasoning: A Training-Free Protocol for Exploiting Answer-Space Constraints in Mathematical Problem Solving cs.CL | cs.SCPDF

Hongbo Ma, Bangji Yang, Yunqian Selina Cheng, Jiajun Fan, Hanwen Zhang

TL;DR: 本文提出了一种名为约束优先推理（CFR）的训练免费两阶段提示协议，用于解决大语言模型在数学问题求解中违反显式约束的问题。该方法首先提取并总结问题中的约束条件，然后在求解过程中检查中间和最终结果是否符合这些约束。通过路由机制，仅在检测到限制性提示时激活两阶段协议，否则使用直接思维链推理。

Details

Motivation: 大语言模型在生成数学对象时经常违反显式要求，例如忽略模运算、返回非整数或使用错误的编码答案形式，这促使研究者开发一种无需训练的方法来利用答案空间约束以提高求解准确性。

Result: 在AIME、CMIMC、BRUMO和AIMO_AMC等多个数学竞赛基准测试中，该方法在多个骨干模型上改进了直接思维链推理的性能，并通过实验验证了其解码鲁棒性和约束质量。

Insight: 创新点在于将约束提取与求解过程分离，通过两阶段协议实现针对性的测试时干预，而非通用数学推理替代方案；其有效性依赖于可恢复的约束和可靠的第一阶段提取，为利用问题特定约束提供了结构化方法。

Abstract: Large language models can derive a plausible mathematical object yet still violate explicit requirements–for example, by omitting a modular reduction, returning a non-integer, or using the wrong encoded answer form. We introduce Constraint-First Reasoning (CFR), a training-free two-stage prompting protocol: Stage 1 extracts and summarizes constraints entailed by the problem, and Stage 2 solves while checking intermediate and final results against that summary. Routed-CFR activates the two-stage protocol only when a text-only regex router detects restrictive cues; otherwise it uses direct chain-of-thought (CoT). Across AIME, CMIMC, BRUMO, and AIMO_AMC, the method improves direct CoT on multiple backbones. We further report convention-controlled routing experiments, matched prompting baselines, problem-level paired tests, decoding robustness, constraint-quality audits, total-token accounting, and an OlympiadBench evaluation. These analyses position CFR as a targeted test-time intervention whose benefit depends on recoverable constraints and reliable Stage 1 extraction, rather than as a general-purpose replacement for mathematical reasoning.

[10] Evidence Lock Before Commitment: A Frozen Interface Degrades LLM-as-Judge Evaluation cs.CLPDF

Divyansh Singh

TL;DR: 本文研究了LLM作为评委（LLM-as-Judge）评估中证据锁定（evidence locking）对判断质量的影响。通过将证据提取和最终裁决分离为多个调用（如两阶段证据锁定或三阶段逐点评分锁定），并与标准单次调用结构化判断进行比较，发现证据锁定会显著降低与人类偏好的一致性并增加答案顺序不一致性。

Details

Motivation: 动机在于检验LLM评委工作流中一个常见假设：将标准提取和证据记录作为中间步骤，是否保留了后续裁决所需的所有信息。研究者质疑这种“证据先于承诺”的接口设计可能因冻结的中间表示而损害最终判断质量。

Result: 在HelpSteer3、FeedbackQA和CoVal三个数据集上进行了总计24,000次判断实验，使用Claude Sonnet 4.5和GPT-5作为评委。结果显示，与结构化单次调用判断相比，两阶段证据锁定使与已发布人类偏好的一致性降低了4-6个百分点，并使答案顺序不一致性增加了8-10个百分点。三阶段逐点评分锁定同样有害，而结构化证据提取则与标准判断结果接近。

Insight: 创新点在于通过可控实验揭示了LLM评委评估流程中接口设计的关键影响：将证据持久化并作为后续裁决的唯一输入（证据锁定）会显著降低判断质量。核心洞察是，持久化证据有助于可审计性，但在决策时不应完全替代原始答案作为输入，这为设计更可靠的LLM-as-Judge系统提供了重要指导。

Abstract: LLM judges are often asked to extract criteria and evidence before choosing between candidate answers. This workflow assumes that the intermediate record preserves the information needed for a later verdict. For reasoning-capable models, visible field order does not reveal internal decision order, so we test an observable alternative: persist the evidence in one call and make it the exclusive input to the next. Across 24,000 judgments over HelpSteer3, FeedbackQA, and CoVal, we compare standard pairwise judging, structured one-call judging, two-call evidence locking, and three-call pointwise locking with Claude Sonnet 4.5 and GPT-5. Evidence locking reduces agreement with released human preferences by 4 to 6 percentage points and increases answer-order inconsistency by 8 to 10 points relative to structured one-call judging. Pointwise locking is also harmful, while structured evidence elicitation remains close to standard judging. The result holds for both judges and all three datasets. Persisted evidence can support auditability, but it should not replace the source answers at decision time.

[11] Answer First, Reason Later: Commitment Order in Diffusion LLMs cs.CL | cs.AIPDF

Jewon Yeom, Jaewon Sok, Seonghyeon Park, Jeongjae Park, Hwiyeong Lee

TL;DR: 本文研究了掩码扩散语言模型（dLLMs）在推理任务中的解码顺序问题，发现其任意顺序生成令牌的自由度反而导致模型过早输出答案而忽略推理过程。通过分析LLaDA-8B在GSM8K上的解码轨迹，作者揭示了这种‘先答后推’的崩溃现象，并提出了一种名为‘前沿门控承诺’的干预方法，在保持部分并行解码能力的同时显著提升推理性能。

Details

Motivation: 旨在探究扩散语言模型在推理任务中因令牌生成顺序不受约束而导致的性能下降问题，特别是模型倾向于过早输出最终答案而跳过中间推理步骤的病理现象。

Result: 在GSM8K和MATH-500等数学推理基准上，无约束解码导致高达90%的问题仅生成答案而无推理过程；提出的‘前沿门控承诺’方法将性能从0.528显著提升至0.852，同时保留了最高4倍的并行解码速度。

Insight: 核心创新在于揭示了扩散模型在推理任务中失败的根本机制是‘可达性’问题而非终止信念，并提出了一个简单的单参数干预方法；该研究将现有基于窗口的采样器重新阐释为针对此推理病理的最小修复方案，而非仅出于效率动机。

Abstract: Masked diffusion language models (dLLMs) can commit tokens in any order – a freedom marketed as their core advantage over autoregressive decoding. We show that on reasoning tasks this freedom is instead the axis of failure. Logging every commitment during decoding of LLaDA-8B on GSM8K, we find that unconstrained (pure) decoding commits the final answer at 15-24% of the trajectory while half the reasoning region is still masked, and collapses to answer-only outputs on up to 90% of problems as the canvas grows. The cause is not the model’s termination beliefs – EOS “pressure” is nearly identical across decoders – but reachability: whether the sampler may act on those beliefs at distant positions. A 2x2 prompt-decoder design shows that chain-of-thought helps only under ordered commitment (interaction +34.8 percentage points, 95% CI [26.8, 42.8]; without reasoning text the decoders are indistinguishable), an interaction we decompose into a collapse channel and an order channel and replicate on Dream-7B and MATH-500. A single-knob intervention – frontier-gated commitment – causally recovers the full gap (0.528 to 0.852) while preserving up to 4x parallel decoding, along a measured frontier whose optimal window flips from w=1 at full refinement to unconstrained at 8 tokens/step. Our results reframe existing window-style samplers, previously motivated by efficiency, as the minimal fix for a reasoning pathology they were never designed to address.

[12] On-Policy Delta Distillation for Multilingual Math Reasoning cs.CL | cs.LGPDF

Byeongho Heo, Jaehui Hwang, Sangdoo Yun, Dongyoon Han

TL;DR: 本文研究了在策略蒸馏（OPD）及其改进版本在策略增量蒸馏（OPD²）在多语言数学推理任务中的应用，特别是在英语、韩语和日语上。实验表明，OPD²通过利用后训练教师模型与其基础模型之间的概率差距作为学习信号，能持续超越原始OPD，显著提升韩语和日语性能，并缩小英语与韩语之间的性能差距。

Details

Motivation: 在策略蒸馏（OPD）作为大语言模型后训练的一种有前景的替代强化学习的方法，但其在多语言环境下的有效性尚未充分探索。本文旨在研究OPD及其变体OPD²在多语言数学推理中的表现，以解决多语言场景下的性能优化问题。

Result: 基于Qwen3模型的实验结果显示，OPD²在英语、韩语和日语数学推理上均优于原始OPD，尤其在韩语和日语上提升显著，并普遍缩小了英语与韩语之间的性能差距。同时发现仅使用英语数据的OPD也能提升韩语和日语性能，但可能导致回答偏向英语。

Insight: 论文的创新点在于提出了OPD²方法，利用教师模型与基础模型的概率差作为更精细的学习信号，以提升多语言推理性能。从客观角度看，这提供了一种高效的多语言后训练策略，并强调了使用多语言数据对于保持目标语言响应特性的重要性，对跨语言模型优化具有借鉴意义。

Abstract: On-Policy Distillation (OPD) is emerging as a promising alternative to reinforcement learning for LLM post-training, yet its effectiveness in multilingual settings remains underexplored. We study OPD and its advanced variant, On-Policy Delta Distillation (OPD$^2$), for mathematical reasoning in English, Korean, and Japanese. OPD$^2$ improves OPD by using the probability gap between a post-trained teacher and its base model as the learning signal. Experiments with Qwen3 show that OPD$^2$ consistently outperforms the original OPD, with particularly strong improvements in Korean and Japanese, and generally narrows the English-Korean performance gap. We further find that English-only OPD can also increase performance for Korean and Japanese, but often shifts the responses toward English, highlighting the importance of multilingual data to preserving target-language responses.

[13] Hierarchical Latent Prediction for Language Models cs.CL | cs.AIPDF

Chang Shi, Tim Pearce, Manan Tomar, Siddhartha Sen, John Langford

TL;DR: 本文提出了一种名为分层潜在预测（HiLP）的新方法，用于改进语言模型的预训练。该方法通过引入更高层次的抽象潜在变量来辅助预测，旨在缓解标准下一词预测（NTP）在长程推理和规划中的局限性，以及现有方法如多词预测（MTP）和下一潜在预测（NextLat）中存在的视野有限或误差累积问题。

Details

Motivation: 标准下一词预测（NTP）的教师强制训练范式可能不适用于长程推理和规划，而现有改进方法（如MTP和NextLat）存在预测视野有限或多步展开时误差累积的问题。

Result: 实验表明，HiLP能产生更长程、更连贯的信念状态表示，在编码和多步推理基准测试中证明了其有效性，并提供了更高的推测解码效率。

Insight: 核心创新点在于引入了分层的、更高层次的抽象潜在变量来辅助潜在空间中的多步展开预测，这有助于减少误差累积效应，从而提升模型在长程任务中的表现和解码效率。

Abstract: While standard Next-Token Prediction (NTP) lays the foundation of language model pre- training, its teacher-forced training paradigm may not be optimal for long-horizon reasoning and planning. Recent works such as Multi-Token Prediction (MTP) and Next-Latent prediction (NextLat) try to mitigate the problem through predicting multiple future tokens and self-supervised prediction in the latent space. However, those auxiliary objectives either have a limited horizon or suffer from compounding error from multi-step rollout. We introduce Hierarchical Latent Prediction (HiLP), which introduces an auxiliary higher-level abstract latent to help reduce the error accumulation effect in latent-space rollouts. Experiments show that HiLP can lead to longer-horizon coherent belief state representation and demonstrate the effectiveness of our method across coding and multi-step reasoning benchmarks, and offers more speculative decoding efficiency.

[14] M$^3$R-Bench: A Unified Benchmark for Evidence-Grounded Multimodal Metaphor Understanding cs.CLPDF

Hong Jiang, Junnan Zhu, Jingwang Huang, Xiao Sun, Yuming Yang

TL;DR: 该论文提出了M³R-Bench，一个用于评估证据驱动的多模态隐喻理解能力的统一基准数据集，包含1000个人工标注的图像-文本实例。为了解决现有模型在跨模态推理中忽略视觉证据、依赖浅层文本线索的问题，作者还提出了M³R-Reasoner模型，该模型结合了课程式推理监督和任务感知强化学习，以提升隐喻解释的准确性。

Details

Motivation: 现有基准主要通过孤立子任务评估隐喻理解，缺乏基于证据的解释，难以评估模型是否基于视觉和文本线索建立了正确的映射关系。因此，需要一个新的、统一的、基于证据的基准来全面评估多模态隐喻理解。

Result: 在M³R-Bench上的实验表明，M³R-Reasoner（仅使用80亿参数主干）在四个统一任务指标上超越了更大的专有多模态大语言模型，并且在视觉证据和情感合理性得分上分别比GPT-5.5高出28.45和30.11分，同时在平均评分上超过Claude-Sonnet-4.6达8.00分。

Insight: 创新点在于构建了一个基于概念隐喻理论和非字面语言理解理论的统一基准，并提出了结合课程式推理监督与任务感知强化学习的模型，以解决跨模态证据与映射不匹配的问题，从而更有效地对齐模型推理与隐喻解释过程。

Abstract: Metaphor enables the understanding of abstract concepts through cross-domain mappings while conveying affective attitudes. In multimodal scenarios, visual and textual information jointly construct Target–Source mappings, requiring both conceptual understanding and cross-modal reasoning. However, existing benchmarks mainly evaluate metaphor understanding through isolated subtasks and lack evidence-grounded explanations, making it difficult to assess whether models establish mappings grounded in visual and textual cues.To address these limitations, we introduce M$^3$R-Bench, a unified and evidence-grounded benchmark containing 1,000 image–text instances with human-verified annotations. Guided by Conceptual Metaphor Theory and theories of nonliteral language understanding, M$^3$R-Bench provides joint annotations for metaphor occurrence, Target–Source mapping, sentiment, and stage-wise explanations following ``evidence identification–mapping establishment–sentiment inference.’’Evaluations on M$^3$R-Bench reveal that existing models often overlook visual evidence, rely on superficial textual cues, and produce inaccurate Target–Source mappings, exposing a cross-modal evidence–mapping mismatch. To address this mismatch, we propose M$^3$R-Reasoner, which combines curriculum-based reasoning supervision with task-aware reinforcement learning to align model reasoning with metaphor interpretation. Experiments show that, with only an 8B-parameter backbone, M$^3$R-Reasoner outperforms larger proprietary MLLMs across four unified-task metrics and improves Visual Evidence and Sentiment Justification scores over GPT-5.5 by 28.45 and 30.11 points, respectively, while surpassing Claude-Sonnet-4.6 by 8.00 points in mean rubric score. The dataset and code are available at https://github.com/hongshi4/M3R-Bench.

Wenhao Xu, Kaiwen Zhang, Hao Li, Maowei You, Yongzheng Ji

TL;DR: 本文提出了隐式社交情境分析（MoCA）任务，旨在系统建模隐式社交场景中的情感、意图和立场三个维度。作者构建了一个包含3,108个多模态实例的高质量基准数据集，并提出了冲突驱动溯因推理（CoDAR）框架来提升模型对隐式社交信息的推理能力。实验表明，现有先进多模态大语言模型在该任务上表现不佳，而CoDAR能显著提升性能，但仍与人类推理存在较大差距。

Details

Motivation: 现实社交沟通中，情感和意图等常通过隐式、间接的方式传达，缺乏系统研究框架。本文旨在填补这一空白，系统建模隐式社交情境。

Result: 在MoCA基准测试中，现有SOTA多模态大语言模型表现显著不足；提出的CoDAR框架大幅提升了模型性能，但仍远低于人类水平。

Insight: 创新点包括：1) 提出MoCA任务，系统定义隐式社交情境的三个维度；2) 构建高质量多模态基准数据集，含细粒度认知标注；3) 提出CoDAR框架，通过建模观察表达与真实行为间的认知冲突来推理隐式心理状态。

Abstract: Human social communication, such as affection and intent, is often conveyed in highly implicit ways, where underlying meanings are expressed through indirect, socially and culturally grounded signals rather than explicit statements. Such implicit social contexts are pervasive in real-world interactions, yet there remains a lack of a formal and systematic framework for studying them. In this paper, we introduce Implicit Social Context Analysis (MoCA), a novel task that systematically models implicit social scenarios along three key dimensions: affection, intent, and stance. We construct a high-quality benchmark containing 3,108 multimodal instances collected from real-world sources, with fine-grained cognitive annotations revealing who expresses what toward whom, as well as how and why it is conveyed. Using the MoCA dataset, we show that state-of-the-art multimodal large language models struggle significantly with this task because of their reliance on explicit cues and limited ability to reason over latent social contexts. To address this challenge, we propose Conflict-Driven Abductive Reasoning (CoDAR), a novel framework that models the discrepancy between observed expressions and expected truthful behavior as cognitive conflict, thereby enabling the inference of hidden mental states. Extensive experiments demonstrate that CoDAR substantially improves model performance. Nevertheless, a large gap from human reasoning remains, highlighting the fundamental difficulty of implicit social understanding.

Xiaofeng Wang, Kakam Chong, Shuai Xiao, DeXin Kong, Qingyuan Tian

TL;DR: 本文提出了一种名为Think-Strategy-Response（TSR）的层次化推理框架，旨在增强大语言模型在动态社交互动中的社会智能。该框架将社交对话分解为高层策略规划和低层语言执行两个阶段，并引入了结合方差门控奖励的线性化层次强化学习算法（LHRL-VGR）进行优化。在SOTOPIA基准测试中，该方法微调的Qwen2.5-7B智能体在目标完成成功率上超越了GPT-4o基线7.32%，实现了最先进的性能。

Details

Motivation: 当前大语言模型在结构化任务上表现出色，但在需要长期目标协调和快速适应的动态社交互动中存在困难。现有方法通常对每个话语应用统一的目标奖励，忽略了每个对话轮次目标的特异性，也未能考虑潜在策略的合理性。

Result: 在SOTOPIA基准测试中，该方法微调的Qwen2.5-7B智能体在目标完成成功率上超越了GPT-4o基线7.32%，在多智能体社交谈判任务中实现了最先进的性能。

Insight: 核心创新点在于受计划行为理论启发，提出了TSR层次化对话框架，以及LHRL-VGR算法。该算法通过基于目标达成分数方差的动态奖励路由，平衡目标完成与策略遵循，实现了更精细的对话行为优化。

Abstract: Large language models (LLMs) excel in structured tasks but struggle with dynamic social interactions, where success requires long-term goal coordination and rapid adaptation. Current methods often apply uniform goal-based rewards to every utterance, overlooking the specificity of objectives at each dialogue turn and failing to account for the rationale of potential strategies. Inspired by the Theory of Planned Behavior, we propose the Think-Strategy-Response (TSR) framework, which decomposes social dialogue into two hierarchical stages: high-level strategic planning and low-level linguistic execution. To optimize TSR, we introduce Linearized Hierarchical Reinforcement Learning with Variance-Gated Rewards (LHRL-VGR), a novel algorithm that dynamically routes rewards - balancing goal completion and strategy adherence - based on the variance of goal achievement scores. Experiments on the SOTOPIA benchmark show that our approach fine-tunes a Qwen2.5-7B agent to surpass the GPT-4o baseline by 7.32% in goal completion success, demonstrating state-of-the-art performance in multi-agent social negotiation tasks.

[17] MACRO: Markov Chain Routing of Transformer Layers cs.CL | cs.AIPDF

Paweł Batorski, Abtin Pourhadi, Akylgali Aitaza, Przemysław Spurek, Paul Swoboda

TL;DR: 本文提出了MACRO框架，用于在大型语言模型（LLM）中实现动态层路由，而无需修改模型权重。该方法将层路由建模为上下文相关的马尔可夫策略，支持跳过、重复和残差隐藏状态加法等操作，并通过训练数据反馈进行更新。在多个推理和知识基准测试中，MACRO显著提升了模型性能，尤其是在小型模型上，并大幅减少了路由搜索时间。

Details

Motivation: 标准的LLM按顺序执行各层，而动态层路由（如层重复、跳过等）可以提升性能。现有路由方法通常需要更新模型权重、为每个测试实例运行昂贵的搜索循环，或在推理时需要真实标签。本文旨在解决这些问题，学习任务特定的路由路径而不修改底层参数。

Result: 在多个开源LLM的多样化推理和知识基准测试中，MACRO相比未路由的基线平均准确率提升了+5.0%，在小型模型上提升最大。它比最佳动态路由方法Dr. LLM高出+7.2%，同时将路由搜索时间减少了9.4倍（从14.8小时降至1.6小时）。

Insight: 创新点在于将层路由建模为基于层索引、计算预算阶段、方向位移和算子上下文的马尔可夫策略，并通过top-k Viterbi算法解码高概率候选程序。这提供了一种无需修改权重、高效且任务自适应的动态路由框架，显著提升了推理效率与性能。

Abstract: Standard Large Language Models (LLMs) execute layers sequentially. Dynamic layer routing, i.e. search for a different execution path through layers involving layer repetitions, skips and other moves, can improve performance. Existing routing approaches often require updating model weights, running expensive search loops per test instance, or demand ground-truth labels during inference. In this work, we propose Markov Chain Routing of Transformer Layers (MACRO), a framework that learns task-specific routes over LLM architectures without modifying underlying parameters. MACRO models layer routing as a context-dependent Markov policy conditioned on layer indices, computation budget phases, directional displacements, and operator context, supporting skip, repeat, and residual hidden-state addition operations. The Markov route distribution is updated via feedback on training data and decoded using a top-k Viterbi algorithm to isolate high-probability candidate programs. We evaluate MACRO across diverse reasoning and knowledge benchmarks on multiple open-weight LLMs. MACRO achieves a +5.0% average accuracy improvement over the unrouted baselines, with largest gains on small models. We outperform the best dynamic routing approach Dr. LLM by +7.2%, while reducing route-search time 9.4x (from 14.8 to 1.6 hours). Our code is publicly available at https://github.com/Batorskq/MACRO.

[18] Causal Episodic Memory for Feedback-Driven Agent Repair cs.CLPDF

Khang Nhat Hoang Vo, Tam Minh Chu, Anh Trac Duc Dinh, Thuyen Vinh Ha Bui, Tho Quan

TL;DR: 本文提出了MERIT，一种无需训练的LLM智能体，通过维护在线双极性记忆（记录已验证的正确修正和观察到的失败方向）来改进Text-to-SQL任务中的失败修复过程。在Spider和BIRD基准测试中，该方法在保持相同初始预测和修复预算的条件下，相比无状态的迭代修复提升了执行准确率。

Details

Motivation: 现有LLM智能体在修复失败时常常丢弃成功的修正，导致后续任务需要重新发现类似解决方案，本文旨在研究如何利用已完成的修复结果来改进后续的Text-to-SQL任务，而无需进行参数更新。

Result: 在Spider基准上，执行准确率从66.34%提升至69.79%；在BIRD基准上，从47.35%提升至48.44%。与Reflexion风格的内存方法相比，MERIT在BIRD上达到51.24%的准确率但推理成本显著更高，且与无类型动态检索方法在统计上未可靠分离。

Insight: 创新点在于引入了基于因果跨查询记忆的在线双极性记忆机制，并通过确定性分类器分配粗略失败类型，以条件化混合词汇-稠密检索器。消融实验表明，负面记忆贡献有限，类型条件化和检索策略的价值依赖于数据集，而模式局部经验提供了最一致的收益。

Abstract: LLM agents that repair failures often discard successful corrections, forcing later episodes to rediscover similar solutions. We study whether finalized repair outcomes can improve subsequent Text-to-SQL episodes without parameter updates. We introduce MERIT, a training-free agent that maintains an online dual-polarity memory of oracle-verified corrections and observed unsuccessful directions. Under oracle-assisted benchmark feedback, only memories from earlier finalized episodes are eligible for retrieval. A deterministic classifier assigns a coarse failure type, which conditions a hybrid lexical-dense retriever before the frozen model generates each revision. Using Qwen2.5-7B-Instruct with identical initial predictions and repair budgets, MERIT improves execution accuracy over stateless iterative repair from (66.34%) to (69.79%) on Spider and from (47.35%) to (48.44%) on BIRD. Paired analyses provide clear evidence for the Spider gain but weaker evidence on BIRD. MERIT is not reliably separated from untyped dynamic retrieval on either benchmark, while Reflexion-style memory reaches (51.24%) on BIRD at substantially higher inference cost. Ablations show that negative memory contributes modestly, the value of type conditioning and lexical–dense ranking is dataset dependent, and schema-local experience provides the most consistent benefit. These results clarify when causal cross-query memory improves repair and when broader memory representations remain preferable.

[19] EpiBench: Can LLMs Understand Epitopes for Antibody Drug Discovery? cs.CL | q-bio.GNPDF

Zirui Wang, Jiaqi Wang, Qinghan Wang, Yuzhi Xu, Gang Du

TL;DR: 本文提出了EpiBench，一个用于评估大语言模型在抗体药物发现中表位理解能力的基准测试。该基准包含1,609个基于结构、功能和突变扫描数据的样本，涵盖五个关联任务，旨在测试模型直接从抗原和抗体序列推断表位信息的能力。

Details

Motivation: 现有表位资源通常专注于孤立的预测任务或依赖专门的结构设置，而通用的蛋白质基准测试未能评估抗体开发流程中以表位为中心的决策。因此，需要建立一个基于序列、可自动评分的基准来填补这一空白。

Result: 对九个通用大语言模型的评估表明，当前模型能够捕捉部分表位相关信号，但在抗体特异性序列基础、长上下文残基定位和基于生物学的推理方面仍存在局限。

Insight: 创新点在于构建了一个闭环、基于序列且自动可评分的诊断性测试平台，通过五个相互关联的任务系统地评估模型对表位的理解，有助于推动可靠的LLM辅助抗体发现。

Abstract: Epitopes determine where antibodies bind antigens and shape downstream therapeutic properties such as functional blockade and escape resistance, making epitope understanding central to antibody drug discovery. Although large language models (LLMs) have shown strong biomedical reasoning ability, it remains unclear whether they can infer epitope information directly from antigen and antibody sequences. Existing epitope resources typically focus on isolated prediction tasks or rely on specialized structural settings, while general protein benchmarks do not evaluate epitope-centered decisions across the antibody development workflow. To address this gap, we introduce EpiBench, a closed-book, sequence-based, and automatically scorable benchmark for evaluating epitope reasoning in LLMs. EpiBench contains 1,609 curated samples grounded in structural antibody–antigen contacts, curated functional B-cell assays, and deep mutational scanning escape measurements. It covers five connected tasks: targetable region discovery, antibody-conditioned epitope identification, epitope binning, functional epitope assessment, and antibody escape assessment, with controlled sampling to reduce shortcut-based evaluation artifacts. We evaluate nine general-purpose LLMs and analyze their behavior through task-specific baselines, antigen length stratification, explicit-reasoning comparison, and failure-mode inspection. The results show that current LLMs capture partial epitope-related signals but remain limited in antibody-specific sequence grounding, long-context residue localization, and biologically grounded reasoning. Therefore, EpiBench provides a diagnostic testbed for measuring and improving sequence-aware biomedical LLMs toward reliable LLM-assisted antibody discovery.

[20] Beyond Sequence Order: Syntax-Informed Positional Embeddings for Transformers cs.CL | cs.AIPDF

Haris Riaz, Hyungji Kim, Mihai Surdeanu

TL;DR: 本文提出了语法感知位置嵌入（SiPE），这是一种在Transformer预训练中从依存句法分析中学习轻量级语法先验，并将其注入到绝对、相对和旋转三种主流位置编码家族中的方法。该方法在不改变自注意力机制和模型架构的前提下，提升了模型对句法结构的感知能力。

Details

Motivation: 现有Transformer中的位置编码主要编码词元距离和顺序，但对句法结构不敏感。本文旨在将句法结构信息有效地整合到位置编码中，以提升模型的语言理解和泛化能力。

Result: 在SyntaxGym基准测试上，使用SiPE预训练的模型性能提升了10.3%，同时困惑度降低了9.0%。在GLUE基准测试上，语言理解分数提升了8.2%。这些结果表明SiPE在提升句法泛化能力的同时，也改善了真实世界的语言理解任务。

Insight: 创新点在于将句法先验以轻量级方式注入到位置编码中，且注入方式因架构而异（如解码器中与相对位置项乘性结合，编码器中直接加到输入嵌入）。该方法在保持推理成本不变的前提下，实现了句法监督与模型性能的新帕累托前沿，避免了现有方法在推理时需边缘化多个句法树或丢弃句法信息的缺点。

Abstract: Positional embeddings (PE) in Transformers encode token distance and order but are largely agnostic to \textit{syntactic structure}. We introduce \textbf{S}yntax-\textbf{i}nformed \textbf{P}ositional \textbf{E}mbeddings (\textbf{SiPE}), which learns a lightweight syntactic prior from dependency parses during pretraining and injects it across all three dominant PE families (absolute, relative, rotary), for both encoders and decoders, leaving self-attention and the rest of the architecture untouched. We isolate \emph{where} and \emph{how} the prior should enter the model, and find it depends on the architecture: for autoregressive decoders that use relative PE, the prior is strongest when coupled multiplicatively with the relative-position term of the attention score, outperforming injection into the input embeddings, into self-attention, or into the positional and attention terms jointly—while for encoders it is best added directly to the input embeddings, composing with each encoder’s native positional mechanism. We find that models pre-trained with SiPE improve on the SyntaxGym benchmark by up to $10.3%$ while simultaneously reducing perplexity by $9.0%$ over a base model with no syntactic supervision—a metric nearly every existing syntax-injection method instead degrades. Crucially, these gains extend beyond syntactic generalization: SiPE also improves real-world language understanding, raising scores on the GLUE benchmark by up to $8.2%$ over a model trained without it. Unlike existing syntactic language models that marginalize over many parses at inference or discard syntax at runtime, SiPE conditions on a single parse, establishing a new Pareto frontier between syntactic supervision and inference cost.

[21] NeSy-RAG: Neuro-Symbolic RAG for Explainable Question Answering cs.CL | cs.SCPDF

Jonas Gann, Michael Gertz

TL;DR: NeSy-RAG是一个模块化的神经符号RAG框架，旨在提升问答系统的可解释性。它通过从检索到的文本块中合成可归因的Prolog模块，生成布尔声明谓词并组合成Prolog查询，同时引入符号知识缺口检测机制来处理缺失的用户上下文。执行查询产生确定性答案和透明执行轨迹，将推理步骤与来源关联起来。

Details

Motivation: 解决传统RAG在推理过程不透明、中间步骤难以验证且无法可靠归因于具体证据的问题，以及缺乏系统化检测用户特定上下文缺失导致输出不完整或错误的问题。

Result: 在ShARC基准测试上，无需领域特定训练，NeSy-RAG达到61.1%的准确率，优于相同模型的RAG基线（42.8%准确率），实现了性能提升。

Insight: 创新点在于结合神经与符号方法，通过生成Prolog谓词和查询实现可解释推理，并引入符号知识缺口检测以自动触发交互补充缺失信息，增强了归因性和透明度。

Abstract: Retrieval-augmented generation (RAG) improves question answering by grounding large language models (LLMs) in external knowledge such as text corpora. However, its reasoning process remains largely opaque: intermediate reasoning steps are difficult to verify and cannot be reliably attributed to specific evidence. Moreover, missing user-specific context is rarely detected systematically, often leading to incomplete or incorrect output. We propose NeSy-RAG, a modular neuro-symbolic RAG framework that synthesizes attributable Prolog modules from retrieved text chunks. For each chunk, the system generates semantically meaningful predicates that encode Boolean claims, which may depend on user facts. Using joint natural language-code embeddings, predicates are retrieved and composed into Prolog queries. To address incomplete user context, we introduce a symbolic knowledge-gap detection mechanism that identifies missing user facts whose truth values affect the query outcome and automatically triggers follow-up interactions. Executing the resulting Prolog queries yields deterministic answers together with transparent execution traces that link each reasoning step to its originating source. On the ShARC benchmark, without domain-specific training, NeSy-RAG achieves 61.1% accuracy, outperforming a same-model RAG baseline that achieves 42.8% accuracy.

[22] Routing Is Least Learnable Where It Is Most Valuable: Bounds on Representation Routing for Web Agents cs.CLPDF

Jiaming Wei, Zekun Wu, Adriano Koshiyama, Maria Perez-Ortiz

TL;DR: 该论文研究了网页代理在不同观察模式（文本、像素或两者）下的性能表现，发现不同模式在任务上具有互补性，但为每个任务选择最佳模式的增益有限。论文通过实验表明，由于运行间噪声的存在，简单的固定模式策略在多数情况下与复杂的路由策略效果相当，而路由监督的标签稀缺性限制了其价值。

Details

Motivation: 网页代理通常固定使用单一观察模式，但不同模式在不同任务上表现各异，论文旨在探索为每个任务动态选择最佳观察模式是否能带来显著性能提升。

Result: 在VisualWebArena和WebArena的八个站点-模型组合上，不同观察模式互补，但运行间噪声导致结果变化12-14%。实验测试了五种路由策略，发现除一个稀疏场景外，固定一个精心选择的模式通常与路由策略效果相当，成本节约在9.5-30.6%之间。

Insight: 论文揭示了路由监督的标签稀缺性问题：代理性能越弱，可用于路由的标签越少，而这正是路由最有价值之处。这一限制属于当前代理而非路由本身，更强的代理可能改变结果，同时运行间噪声是评估中需考虑的关键因素。

Abstract: Web agents observe a browser through text, pixels, or both, and the choice is usually fixed once for all tasks. We measure six observation modes across eight site-model combinations (cells) on VisualWebArena and WebArena and ask what choosing per task would buy. The modes are complementary: each solves tasks the others miss, they fail in structurally different ways, and the best choice reverses between task sets. The obvious prize, an oracle that picks a winning mode for every task, looks large but is inflated by run-to-run noise: rerunning the same mode on the same tasks changes 12-14% of outcomes, so a second run of a mode already in hand gains about as much as adding a new one. What survives is a cost bound: sending only the tasks no mode solves to the cheapest mode cuts cost by 9.5-30.6% in 8 of 8 cells at unchanged success. We then test five routing policies (picking the mode, deciding when to spend on the strong mode, a zero-cost rule read off the task text, a confidence cascade, and pooled cost tiers), and none robustly beats simply fixing one well-chosen mode; the one exception is a fragile result in our sparsest cell. The central obstruction is that routing supervision is produced at the agent’s success rate: the weaker the agent, the fewer labels a router gets, exactly where routing would be most valuable. This limit belongs to today’s agents rather than to routing itself. Label supply and routing opportunity rise together (correlation 0.95 across cells), so a stronger agent can overturn the result, and we report the rerun noise bands and the full measurement protocol.

[23] RP-OPSD: Reasoning-Pivot-Guided On-Policy Self-Distillation for Multilingual Reasoning Transfer cs.CLPDF

Xinye Wang, Junxiao Liu, Shujian Huang

TL;DR: 本文提出了RP-OPSD方法，一种基于推理枢纽引导的在线策略自蒸馏方法，用于提升大语言模型在多语言推理任务上的迁移能力。该方法通过识别并优先蒸馏对跨语言推理转移至关重要的‘推理枢纽’信号，在覆盖17种语言的数学推理基准测试上超越了现有基线方法。

Details

Motivation: 现有在线策略自蒸馏方法在为学生模型生成的数据轨迹提供密集的token级监督时，其目标并未明确优先考虑对跨语言迁移最关键的推理信号。本文旨在解决这一不足，通过聚焦于‘推理枢纽’来提升多语言推理迁移效果。

Result: 在覆盖17种语言、多个难度级别的数学推理基准测试上的实验表明，RP-OPSD方法超越了强大的多语言推理基线模型和现有的OPSD变体。

Insight: 核心创新在于将目标语言的推理过程解耦为表面文本生成和‘推理枢纽’生成，并利用教师模型在有/无英文参考解答情况下的分布偏移作为操作代理，来引导特权蒸馏和参考锚定，从而将蒸馏重点集中在推理控制和问题条件状态更新的token上。

Abstract: Multilingual reasoning transfer is crucial for extending reasoning capabilities of large language models (LLMs) beyond high-resource languages. On-policy self-distillation (OPSD) and its variants have emerged as a promising paradigm, providing dense token-level supervision on student-generated rollouts, yet their objectives do not explicitly prioritize reasoning signals most critical to cross-lingual transfer. We characterize that target-language reasoning comprises the generation of both surface text and reasoning pivots, which are decisions that advance or redirect the reasoning process and shape subsequent inference. This motivates concentrating privileged distillation around such pivots. We therefore propose RP-OPSD, Reasoning-Pivot-guided On-Policy Self-Distillation, using the distributional shift between matched teacher views with and without an English reference solution as an operational proxy to guide privileged distillation and reference anchoring. Experiments on mathematical reasoning benchmarks covering 17 languages and multiple difficulty levels show that our method outperforms strong multilingual reasoning baselines and OPSD variants. Further analysis reveals that RP-OPSD concentrates privileged distillation on reasoning-control and problem-condistioned state-update tokens, while downweighting it for tokens that mainly support surface realization. Our code is available at https://github.com/NJUNLP/RP-OPSD.

[24] Learning When to Trust via Selective Context Preference Optimization cs.CL | cs.AI | cs.LGPDF

Xian Sun, Wei Chow, Yingshuo Wang, Junhao Liu, Wei Gao

TL;DR: 本文提出了一种新的评估框架MIST和优化方法SCOPE，旨在解决语言模型在依赖外部信号时如何选择性信任的问题。MIST基准包含四种匹配条件（干净、误导、正确上下文和无关上下文），而SCOPE方法通过平衡所有条件的偏好对来优化模型，减少模型被误导信号翻转正确答案的情况。

Details

Motivation: 语言模型在依赖外部信号时，单个误导信号可能导致正确答案被错误翻转，而简单地训练模型抵抗所有信号会使其在值得信任的上下文中变得无用。因此，需要研究模型如何实现选择性信任，而非单纯抵抗。

Result: 在MIST基准上的全面研究表明，模型对误导信号的敏感性是普遍存在的。SCOPE方法显著降低了流行开源模型的SC2W（误导信号翻转正确答案的频率），同时在干净、正确或无关上下文中保持了准确性。

Insight: 创新点在于将问题重新定义为选择性信任，并引入了MIST基准和SC2W配对指标来量化误导影响。SCOPE方法通过挖掘失败案例并平衡所有条件的偏好对进行优化，避免了仅关注误导项的偏差，从而在保持模型有用性的同时提升鲁棒性。

Abstract: Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious remedy, training models to resist such signals, hides a failure mode: a model that ignores all context looks robust yet is useless when the context is worth trusting. We recast the problem as selective trust and introduce MIST, a human-annotated benchmark that renders each reasoning item under four matched conditions (clean, misleading, correct-context, and irrelevant-context), together with SC2W, a paired metric counting how often a misleading signal flips a clean-correct answer to wrong. Across a comprehensive benchmark study, we observe that such a susceptibility is universal. We then propose SCOPE, which mines clean-correct/misleading-wrong failures and optimizes a standard Direct Preference Optimization (DPO) objective over matched preference pairs balanced equally across all four conditions, rather than over misleading items alone. Our approach substantially reduces SC2W on popular open-sourced models while preserving accuracy when the added context is clean, correct, or irrelevant. With this work, we argue that models should be judged on selective trust, not on resistance alone.

cs.CV [Back]

[25] MapTCL: Temporal Consistency Learning via Bidirectional Alignment for Vectorized HD Map Construction cs.CVPDF

Hyeonseo Kim, Juyeb Shin, Hyeonjun Jeong, Hiwon Shin, Dongsuk Kum

TL;DR: 论文提出了MapTCL，一种用于提升在线高精地图构建时序一致性的辅助训练策略。该方法通过双向对齐设计了两种损失函数：BVCL用于建模关联向量实例间的几何与语义差异，RCL用于稳定稠密BEV特征。联合训练后，MapTCL能有效提升生成地图的时序稳定性，且作为即插即用模块可增强现有基线模型。

Details

Motivation: 在动态城市环境中，移动物体和遮挡使得构建可靠的在线高精地图具有挑战性。现有基于特征级时序融合的方法仅依赖逐帧真值监督，缺乏直接惩罚连续地图间几何噪声和时序抖动的显式目标。

Result: 在nuScenes和Argoverse 2两个标准基准测试上的大量实验表明，MapTCL作为通用即插即用模块，能持续提升现有基线模型性能：在nuScenes上获得+3.7 mAP和+2.8 C-mAP的提升，在Argoverse 2上获得+3.1 mAP和+2.5 C-mAP的提升，且无需额外推理开销。

Insight: 创新点在于提出了显式的时序一致性学习框架，通过双向向量一致性学习和栅格图一致性学习双损失联合优化，直接约束地图的时序平滑性。该方法设计为辅助训练策略，不增加推理成本，具有良好的通用性和可扩展性。

Abstract: Constructing reliable online HD maps remains challenging in dynamic urban environments due to moving objects and occlusions. While recent works employ feature-level temporal fusion to address this, they rely solely on per-frame ground truth supervision. Consequently, they lack an explicit objective to directly penalize the geometric noise and temporal jitter between consecutive online HD maps. To address this, we propose MapTCL, an auxiliary training strategy that formulates temporal consistency loss between current and past frames via bidirectional alignment. Specifically, Bidirectional Vector Consistency Learning (BVCL) models the geometric and semantic discrepancies between associated past and current vector instances as an auxiliary loss. We also employ Raster map Consistency Learning (RCL) as an additional loss to stabilize dense BEV features. By jointly training with these dual losses, MapTCL improves the temporal stability of generated HD maps. Extensive experiments on two standard benchmarks demonstrate the effectiveness of our approach. As a versatile plug-and-play module, MapTCL consistently enhances existing baseline models, achieving gains of +3.7 mAP & +2.8 C-mAP on nuScenes and +3.1 mAP & +2.5 C-mAP on Argoverse 2 without additional inference overhead.

[26] Innocent Panels, Hateful Stories: Evaluating and Detecting Hateful Intent in Multi-Turn Visual Story Generation cs.CV | cs.AI | cs.CRPDF

Ye Leng, Junjie Chu, Yiting Qu, Mingjie Li, Yun Shen

TL;DR: 本文研究了前沿文本到图像（T2I）系统在生成连贯视觉故事时可能传播仇恨叙事的问题。作者构建了HatefulStoryPrompts数据集评估模型生成仇恨故事的能力，并创建了HatefulVisualStory数据集评估现有审核系统的缺陷。论文提出了互补的主动和生成后防御方法，以检测和缓解多轮视觉故事生成中的群体级仇恨意图。

Details

Motivation: 现有的T2I系统安全研究主要关注单张图像的仇恨内容生成，而忽略了跨多轮对话、由有序图像组共同传达的群体级仇恨叙事。随着T2I系统能够生成具有一致角色和场景的连贯视觉故事，这种可扩展的仇恨内容传播成为一个新的安全威胁，需要被系统性地评估和防御。

Result: 在HatefulStoryPrompts上的评估显示，五个前沿模型都能完成超过80%的仇恨故事，最强的模型达到99.0%。在HatefulVisualStory数据集上，现有审核系统表现不佳：专用安全模型召回率最高仅34.9%，而一个强大的视觉语言模型达到67.5%。论文提出的防御方法中，交互感知监控器在仅提示的会话中召回率达97.3%，在用户提供首张图像时达92.6%；生成后联合分析方法召回率达80.2%。

Insight: 论文的核心创新在于将T2I安全研究从单张图像层面扩展到多轮、连贯的视觉叙事层面，揭示了群体级仇恨含义是现有审核系统的盲点。提出的防御框架（包括交互感知监控和生成后联合分析）为应对这一新威胁提供了技术路径，强调了安全机制需要从单图像审核演进到对交互状态和图像关系的状态化推理。

Abstract: Picture books and comics have long been used to disseminate hateful narratives because they are easily understood even by children, as exemplified by the notorious Nazi propaganda picture book \emph{Der Giftpilz}. Recently, frontier text-to-image (T2I) systems such as Gemini and GPT-Image have enabled conversational generation with consistent characters and scenes across turns, making hateful visual stories, namely ordered image groups that collectively convey hateful narratives, cheap and scalable to produce. Although prior work has studied hateful content generation by T2I systems, it focuses on individual images, leaving group-level hateful meaning largely unexplored. We aim to address the gap. Concretely, we introduce \texttt{HatefulStoryPrompts}, comprising 330 multi-turn configurations from 55 hateful stories across two languages and three visual styles, and evaluate five frontier models over 4,950 attempts. Every model completes over 80% of the stories, with the strongest reaching 99.0%. We further evaluate existing moderation systems on \texttt{HatefulVisualStory}, a human-labeled dataset of 969 hateful image sets and 990 benign controls, and find that they frequently miss group-level hateful meaning: dedicated safety models achieve at most 34.9% recall, while a strong vision-language model reaches 67.5%. Finally, we propose complementary proactive and post-generation defenses. An interaction-aware monitor achieves 97.3% recall for prompt-only sessions and 92.6% when the user supplies the first image, while post-generation methods jointly analyzing completed image groups reach 80.2%. Our work shows that, as image generation evolves from isolated outputs to coherent visual narratives, safety must evolve accordingly, from per-image moderation to stateful reasoning over interactions and image relationships.

[27] Grad-CAM for Vision Transformers: A Systematic Taxonomy and Audit of Methodological Ambiguity in Explainable AI cs.CVPDF

Casey Wall, Longwei Wang, Rodrigue Rizk, KC Santosh

TL;DR: 本文对Grad-CAM方法在Vision Transformers上的应用进行了系统性分类与文献审计，发现大多数论文未清晰说明如何将Grad-CAM适配到Transformer架构，并提出了一个描述性分类法来明确这些被隐式处理的方法学选择。

Details

Motivation: Grad-CAM最初为卷积神经网络设计，其空间特征图和通道维度具有明确的架构意义，而Vision Transformers通过token、注意力、残差流和多模态交互表示图像，结构不同。论文旨在系统梳理和审计Grad-CAM及相关方法在ViT架构上的适配、论证和报告方式，以揭示方法学上的模糊性。

Result: 通过对超过550篇论文的初步搜索，识别出175篇将Grad-CAM或类似方法应用于ViTs的论文。研究发现，大多数论文未提供完整的数学或实现层面的适配说明，表明Grad-CAM在ViTs上的应用常被视为CNN版本的简单扩展，而忽略了影响严谨性、可复现性和解释性的关键选择。

Insight: 论文的创新点在于提出了一个描述性分类法，明确ViT Grad-CAM适配中的特征位置、梯度目标、空间重建步骤和聚合选择，这有助于澄清方法学选择的范围，提升可解释AI的严谨性和透明度。从客观角度看，该研究揭示了可解释AI领域在跨架构方法应用中的常见疏漏，为未来研究提供了审计框架和规范化指导。

Abstract: Gradient-weighted Class Activation Mapping (Grad-CAM) is widely used to visualize model decisions, but it was originally formulated for convolutional neural networks, where spatial feature maps and channel dimensions have clear architectural meanings. Vision Transformers (ViTs) do not provide the same structure, instead representing images through tokens, attention, residual streams, and multimodal interactions. This paper presents a systematic taxonomy and literature audit of how Grad-CAM and related methods are adapted, justified, and reported for ViT-based architectures. From an initial search of more than 550 papers, we identify 175 papers that apply Grad-CAM or Grad-CAM-adjacent methods to ViTs. We find that most papers do not provide a full mathematical or implementation-level account of how Grad-CAM is adapted to transformer representations. To characterize this gap, we introduce a descriptive taxonomy of ViT Grad-CAM adaptations that makes explicit the feature locations, gradient targets, spatial reconstruction steps, and aggregation choices that are often left implicit. This taxonomy is not intended to prescribe a single correct adaptation, but to clarify the range of methodological choices being made. The study shows that Grad-CAM on ViTs is often treated as a trivial extension of CNN-based Grad-CAM, despite requiring nontrivial choices that affect rigor, reproducibility, and interpretation.

[28] NeuroAdaptTrainer: A Fiji/ImageJ Plugin for YOLO-Based Neuron Segmentation, InteractiveCorrection and Transfer Learning cs.CVPDF

Daniela Eraso-Casas, Gerard Villarroya-Pique, Esther Serrano-Pertierra, M. Teresa Fernández-Sánchez, Antonello Novellie

TL;DR: 本文介绍了NeuroAdaptTrainer，一个用于Fiji/ImageJ的开源插件，旨在简化神经元显微图像中的计数与分割任务。该插件集成了YOLO实例分割模型，支持自动检测、手动校正，并能通过迁移学习使模型适应新的成像条件，同时包含外部验证模块以量化评估模型性能。

Details

Motivation: 解决神经科学研究中，对神经元培养显微图像进行计数和分割这一常规但耗时任务，传统方法依赖人工或半自动工具，效率低下。

Result: 插件内置外部验证模块，可在保留的标注集上定量比较基础模型与适应后模型的性能，但摘要未提及具体基准测试或SOTA比较结果。

Insight: 创新点在于将深度学习模型无缝集成到专家工作流中，支持交互式校正和迁移学习，降低了非专业用户的使用门槛，同时保持专家监督的核心地位，提升了分割任务的效率和适应性。

Abstract: Neuron counting and segmentation in microscopy images of neuronal cultures is a routine and time-consuming task in neuroscience research, traditionally performed through manual inspection or semi-automatic tools. We present NeuroAdaptTrainer, an open-source Fiji/ImageJ plugin that integrates a YOLO instance-segmentation model directly into the microscopist’s workflow. The plugin allows a user to run automatic neuron detection on a single image or a batch of images, manually correct the resulting detections from within Fiji, and use those corrections to adapt the model to new imaging conditions via transfer learning. A built-in external validation module allows the base and adapted models to be compared quantitatively on a held-out annotated set. NeuroAdaptTrainer lowers the barrier for non-specialist users to benefit from deep-learning-based segmentation while keeping expert supervision at the center of the workflow.

[29] In-Context Forcing: Uncovering Context Effects in Autoregressive Video Diffusion cs.CV | cs.AIPDF

Lingxiao Yang, Liu Liu, Moran Li, Han Feng, Wenjian Cao

TL;DR: 本文提出了一种名为In-Context Forcing的渐进式自回归范式，用于改进少步自回归视频扩散模型。该方法通过使用噪声水平递减的上下文帧来替代传统方法中完全去噪的干净帧作为参考，从而解决了现有方法因上下文细节泄露导致的时序语义和动态性受损问题。该范式还实现了跨帧并行去噪，显著提升了推理速度。

Details

Motivation: 当前少步自回归视频扩散模型依赖先前完全去噪的干净帧作为上下文，这会导致模型走捷径，泄露过多局部细节，从而损害生成的时序语义和动态性。

Result: 在VBench基准上的大量实验表明，该方法在视觉保真度和推理速度方面均显著优于最先进的方法。

Insight: 核心创新在于提出了一个渐进式噪声上下文机制，即对距离较远的帧施加较少的掩码（噪声），对相邻帧施加较多的掩码，从而提供自适应指导。这解耦了对先前干净帧的严格依赖，实现了跨帧并行去噪，在不牺牲性能的前提下大幅加速推理。

Abstract: Current few-step autoregressive video diffusion models depend on previous fully denoised clean frames as context for all denoising steps of the current frame. However, these clean frames leak excessive local details, which causes the model to take shortcuts, resulting in compromised temporal semantics and dynamics. Inspired by the perspective of diffusion as masking, we explore the impact of noisy contexts on few-step autoregressive generation. Yet, simply applying contexts with the same noise levels provides insufficient guidance, leading to poor temporal consistency. To resolve this dilemma, we introduce In-Context Forcing, a progressive autoregressive paradigm that utilizes contexts with decreasing noise levels. By applying less masking to distant frames and more masking to adjacent ones, this approach provides adaptive guidance, effectively ensuring both robust temporal consistency and high inter-frame dynamics. Furthermore, by decoupling the strict dependence on previous clean frames, our paradigm enables cross-frame parallel denoising, achieving substantial inference acceleration without sacrificing performance. Extensive experiments on VBench demonstrate that our method significantly outperforms state-of-the-art approaches in both visual fidelity and inference speed.

[30] A Paragraph is Worth a Thousand Captions: Rethinking Text Supervision for Vision-Language Retrieval cs.CVPDF

Mahyar Ghazanfari, Amin Tabrizian, Arsyi Aziz, Binshuai Wang, Peng Wei

TL;DR: 本文系统研究了文本监督粒度对视觉-语言检索模型性能的影响，通过合成多样化的标题、困难负样本和质量评分段落数据，在保持视觉编码器冻结的情况下微调BLIP文本编码器。研究发现段落级监督能显著提升长文本检索性能，在ShareGPT4V上与Long-CLIP-L相当，在DOCCI上超越14个百分点，且无需修改模型架构。

Details

Motivation: 现有对比视觉-语言模型（如CLIP、BLIP）主要使用短标题训练，限制了其处理详细文本描述的能力；本文旨在探究文本监督粒度（从单标题到多句段落）是否单独决定长文本检索性能。

Result: 在ShareGPT4V基准上段落监督模型与Long-CLIP-L性能匹配，在DOCCI基准的图像到文本检索中超越Long-CLIP-L超过14个点；段落监督能有效利用长令牌序列（超过60个令牌），而仅标题训练在超过60令牌时性能下降。

Insight: 创新点在于通过控制变量实验（仅微调文本编码器）系统揭示文本粒度对检索性能的影响机制；段落级监督能自然适配长文本输入，且合成数据管道（基于Qwen2-VL和Llama 3.2 Vision）为训练数据构建提供了可扩展方案。

Abstract: Contrastive vision-language models such as CLIP and BLIP are typically trained on short image captions, limiting their ability to retrieve images from detailed textual descriptions. While methods such as Long-CLIP extend the token limit through positional embedding interpolation, we ask a simpler question: does training text granularity alone determine long-text retrieval performance? We present a systematic study of supervision ranging from single captions to multi-sentence paragraphs for contrastive image-text retrieval. Using a synthetic pipeline based on Qwen2-VL and Llama 3.2 Vision, we generate diverse captions, hard negatives, and quality-scored paragraphs for 500K CC3M images. To isolate the effect of text granularity, we fine-tune only the BLIP text encoder while keeping the vision encoder frozen across 10 training configurations. Our paragraph-supervised models match Long-CLIP-L on ShareGPT4V and outperform it by more than 14 points on DOCCI for image-to-text retrieval, without architectural changes. We further show that paragraph supervision enables effective use of long token sequences, whereas caption-only training degrades beyond 60 tokens. Increasing caption diversity improves short-caption retrieval with diminishing returns, while paragraph supervision consistently benefits long-description benchmarks and hard negatives prove detrimental in text-only fine-tuning. Evaluations on Flickr30k, COCO, ShareGPT4V, and DOCCI provide a comprehensive analysis of the trade-offs between text granularity, retrieval direction, and description length.

[31] Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation cs.CV | cs.LGPDF

Yuta Kobayashi, Pradyun Ramesh, Muhammad Ahmed Chaudhry, Vincent Jeanselme, Judy Wawira Gichoya

TL;DR: 本文提出了一种名为PU-DPO的正向-未标记偏好优化框架，用于解决医学影像报告生成中因临床报告存在遗漏噪声（如忽略细微发现）而导致模型学习不足的问题。该框架将缺失的提及视为未标记而非真正的负样本，通过构建对比对来提供偏好监督，从而在存在视觉证据的情况下更倾向于生成提及特定发现的报告。

Details

Motivation: 动机在于解决放射学报告生成中因回顾性临床报告存在遗漏噪声（例如，由于关注设备放置而忽略心脏肥大等细微发现）导致视觉语言模型继承这些遗漏并学习报告不足的问题。

Result: 在半合成实验和真实世界胸部X光基准测试（其中存在已裁决标签）中，PU-DPO在多种病理的检测率和隐藏阳性恢复方面取得了一致的提升，并且比先前方法对遗漏噪声更具鲁棒性。

Insight: 创新点在于将偏好优化重新构建在正向-未标记学习框架下，将缺失提及视为未标记样本，并通过编辑模型响应生成明确提及或省略特定发现的对比对来提供监督，这有助于更准确地捕捉视觉证据并减少遗漏噪声的影响。

Abstract: Vision-Language Models (VLMs) for radiology report generation are typically trained on retrospective clinical reports, which suffer from omission noise: clinically present findings are left unreported due to the omission of subtle findings. For example, prior studies show that cardiomegaly may be omitted from ICU chest X-ray reports when the imaging request is focused on monitoring support device placement. As a result, models trained with standard approaches inherit these omissions, learning to under-report findings themselves. We propose PU-DPO, a preference optimization framework to prevent omission noise from corrupting the preference signal. We reformulate the objective under a positive-unlabeled (PU) learning framework, treating absent mentions as unlabeled rather than truly negative. Our framework provides preference supervision using constructed contrastive pairs, generated using edits to model responses, producing variants that explicitly mention or omit a specific finding. Generated responses that mention the finding are naturally preferred in the context of visual evidence. Across semi-synthetic experiments and analyses on real-world chest radiograph benchmarks where adjudicated labels are available, PU-DPO yields consistent gains in detection rates and recovery of hidden positives across multiple pathologies, and is more robust to omission noise than prior approaches.

[32] LoDA: A Level of Detection Aware Method and a Multimodal Sensing Benchmark for Object Level Change Detection cs.CV | cs.ROPDF

Haitian Wang, Xinyu Wang, Sheldon Fung, Xian Zhang, Zichen Geng

TL;DR: 本文提出了一种面向对象级别的三维变化检测方法LoDA，该方法集成了检测限感知的配准、几何驱动的对象代理以及基于规则的语义和实例分割，并利用高度、体积和表面法向位移线索来分配五种带有置信度的变化标签。同时，作者构建了一个名为LoDA的多模态感知基准数据集，用于评估对象级别的变化检测性能。

Details

Motivation: 现有的从栅格高度差分到深度图像和点云网络的变化检测方法通常是基于图块和阈值驱动的，只能产生逐点分数，缺乏明确的检测限和一致的对象级别标签。因此，需要一种能够可靠检测多时相城市LiDAR中对象级别变化的方法，以保持数字地图与物理世界的一致性。

Result: 在作者构建的Subiaco地区LoDA基准测试中，该方法达到了95.0%的准确率、90.8%的宏观F1分数和83.0%的宏观IoU，比最佳基线高出8.7个IoU点和4.4个F1点。在公开的Urb3DCD-V2基准测试中，按照官方逐点协议评估，达到了96.81%的平均准确率和89.52%的平均变化IoU，分别比已报告的最强基线提高了1.36个mAcc点和3.18个mIoUch点。

Insight: 创新点在于提出了一种将配准、几何和语义解耦的检测流程，能够将姿态不确定性传播到空间变化的检测限中，稳定跨时相的对应关系，并抑制由残余错位和密度变化引起的虚假变化。此外，构建了一个包含多时相车载LiDAR地图、语义实例和对象级别标注的基准数据集，为后续研究提供了评估标准。

Abstract: High-definition 3D LiDAR maps are important for autonomous driving and smart-city services, which require reliable detection of object-level changes in multi-temporal urban LiDAR to keep digital maps aligned with the physical world. Existing approaches from raster height differencing to depth image and point-cloud networks often remain tile-based and threshold-driven, yielding per-point scores without explicit detection limits or consistent object-level labels. We propose an object-level 3D change-detection pipeline that integrates detection-limit-aware registration, geometry-driven object proxies with rule-based semantic and instance segmentation, and displacement cues in height, volume, and surface-normal direction to assign five change labels with confidence. By decoupling registration, geometry, and semantics, the pipeline propagates pose uncertainty into spatially varying detection limits, stabilizes cross-epoch correspondences, and suppresses false changes caused by residual misalignment and density variation. We also present LoDA, a level-of-detection (LoD) aware benchmark for the Subiaco district with fused multi-temporal vehicle-LiDAR maps constructed with LiDAR, GNSS, and IMU support, semantic instances, and object-level annotations. On this benchmark, our method achieves 95.0% accuracy, 90.8% macro F1, and 83.0% macro IoU, exceeding the best baseline by 8.7 IoU points and 4.4 F1 points. On the public Urb3DCD-V2 benchmark evaluated under the official point-wise protocol, it reaches 96.81% mean accuracy and 89.52% mean change IoU, improving over the strongest reported baselines by 1.36 points in mAcc and 3.18 points in mIoUch.

[33] Adapting Vision Foundation Models with Cascaded Semantics cs.CVPDF

Xi Xiao, Xingjian Li, Cheng Han, Tianyang Wang, Lin Zhao

TL;DR: 本文提出了一种名为Cascaded Semantics的视觉提示调优方法，旨在通过注入两种互补的语义先验知识来改进现有的视觉提示调优范式。该方法将基础图像先验（如颜色、纹理、形状）注入输入空间，并将自注意力图提供的实例感知语义注入特征空间，通过级联方案在ViT适应过程中整合这些先验。实验在34个具有挑战性的图像分类数据集上进行，仅调优ViT 0.74%的参数即可实现优异的下游适应性能。

Details

Motivation: 现有视觉提示调优方法中的提示是随机初始化的，未能利用像NLP中指令那样的先验知识，这限制了其适应效率和性能。

Result: 在34个具有挑战性的图像分类数据集上的实验表明，该方法在仅调优0.74% ViT参数的情况下，实现了优异的下游适应性能，达到了SOTA水平。

Insight: 创新点在于将手工提取的基础图像先验和自注意力图提供的实例语义这两种互补先验知识系统地注入并级联整合到视觉提示调优过程中，从而更有效地利用预训练模型的先验知识进行高效适应。

Abstract: Prompt tuning, a leading parameter-efficient adaptation paradigm in NLP, has recently been extended to computer vision. Visual prompt tuning (VPT) adapts pre-trained vision transformers (ViTs) by updating a small set of additional prompt parameters. However, existing visual prompts are randomly initialized and do not exploit prior knowledge, such as instructions in NLP. We address this gap by injecting two complementary semantic priors into VPT. Fundamental image priors, including color, texture, and shape, are extracted with classical hand-crafted operators and injected into the input space, while self-attention maps provide instance-aware semantics in the feature space. We further propose a cascaded scheme that integrates both priors throughout ViT adaptation. Experiments on 34 challenging image classification datasets demonstrate superior downstream adaptation while tuning only 0.74% of ViT parameters. Project page: https://xixiaouab.github.io/Cascaded-Semantics/.

[34] Text-Guided Refinement of Multi-sequence Glioma Subregion Segmentation with a Vision-Language Foundation Model cs.CVPDF

Zach Eidex, Yu-nong Lin, Mojtaba Safari, Sean Pitroda, Ralph Weichselbaum

TL;DR: 本文提出了一种基于三维视觉-语言基础模型（VoxTell）的文本引导脑胶质瘤亚区分割细化框架。该框架利用预训练的VoxTell生成初始分割掩码，并通过可训练的投影层将包含目标、动作、位置等信息的文本提示嵌入注入到其多尺度解码器中，以实现基于文本指令的分割校正。

Details

Motivation: 手动校正脑胶质瘤亚区分割轮廓耗时费力，而现有自动分割模型（如nnU-Net）泛化能力有限，且缺乏基于临床医生文本指令的校正能力。本文旨在探索如何利用三维视觉-语言基础模型实现文本引导的脑肿瘤分割细化。

Result: 在内部测试集（BraTS-GLI）上，使用正确指令将增强后T1加权图像的分割Dice相似系数从0.774提升至0.796，显著优于空白提示和矛盾提示。在跨数据集测试（包括脑膜瘤、转移瘤等）中，正确指令将DSC从0.527提升至0.550，同样优于矛盾指令。

Insight: 创新点在于将视觉-语言基础模型（VoxTell）适配于三维医学图像分割的文本引导细化任务，通过设计包含丰富语义的提示词（Oracle prompts）和轻量化的可训练投影层，实现了对分割结果的指令式、可解释的编辑，而非简单的后处理，为临床医生参与的人机协同工具提供了可能。

Abstract: Background: Accurate glioma subregion delineation is important for radiotherapy planning and longitudinal monitoring, but manual contour correction is time-consuming. Models such as nnU-Net may generalize imperfectly and lack clinician-directed text correction. Purpose: We investigated adapting a three-dimensional (3D) vision-language foundation model for text-guided brain tumor segmentation refinement. Methods: We developed a lightweight VoxTell-based framework. Pretrained VoxTell generated initial masks. Oracle prompts derived from segmentation errors encoded target, action, location, imaging evidence, edit size, and preservation constraints. Frozen Qwen/VoxTell prompt embeddings were injected through trainable projections into its multiscale decoder conditioning; other weights remained frozen. Training, validation, and testing used 901, 100, and 250 BraTS-GLI cases. Cross-dataset transfer was evaluated on 100 meningioma, metastasis, pediatric tumor, and UPENN-GBM cases. Results: On the internal test set using post-contrast T1-weighted input, correct instructions improved subregion Dice similarity coefficient (DSC; enhancing tumor, edema, and necrotic/non-enhancing core) from $0.774\pm0.158$ to $0.796\pm0.137$. They outperformed blank prompts ($0.762\pm0.155$; Holm-adjusted $p<0.001$, $d_z=0.71$) and contradictory prompts ($0.770\pm0.163$; $p<0.001$, $d_z=0.48$). In cross-dataset testing, correct instructions improved DSC from $0.527\pm0.287$ to $0.550\pm0.278$ and outperformed contradictory instructions ($0.504\pm0.275$; $p<0.001$, $d_z=0.43$). Conclusion: A 3D vision-language foundation model can perform instruction-guided refinement of glioma subregion segmentations. Sensitivity to correct, blank, and contradictory prompts suggests text-dependent contour editing rather than nonspecific post-processing, supporting further evaluation as a clinician-in-the-loop tool.

[35] Invisible Shortcuts: Why Vision Encoders Know Your Camera cs.CV | cs.LGPDF

Vladan Stojnić, Ryan Ramos, Giorgos Kordopatis-Zilos, Noa Garcia, Giorgos Tolias

TL;DR: 该论文揭示了深度视觉模型在预训练过程中会学习一种新的捷径：利用图像像素中嵌入的不可见元数据痕迹（如图像处理和拍摄信息）来预测语义标签。作者通过实验证明，大规模语义监督（如ImageNet标签或LAION描述）会自然诱导元数据与语义之间的相关性，使模型将低级信号转换为预测特征。研究还探讨了在预训练期间和之后应用的缓解策略，这些策略能降低模型对元数据的敏感性，同时保持下游任务性能，并可能改善分布外泛化能力。

Details

Motivation: 先前研究主要关注可见的偏差（如物体背景或纹理相关性），而本文旨在识别和探究一种新的捷径学习来源：图像像素层面嵌入的不可见元数据痕迹，这些痕迹与语义监督信号相关，可能导致模型学习到虚假关联。

Result: 通过引入受控的元数据-语义相关性，作者发现更强的相关性会导致模型对元数据痕迹的系统性更高敏感性，以及在元数据分布偏移下更大的性能下降。缓解策略被证明不仅能降低对目标元数据的敏感性，还能泛化到未见过的元数据，且不牺牲下游任务性能。

Insight: 论文的创新点在于首次系统性地识别并验证了视觉编码器对不可见元数据痕迹的敏感性作为一种新的捷径学习形式。从客观角度看，其提出的缓解策略具有借鉴意义，能够提升模型的鲁棒性，并且元数据敏感性本身也部分解释了某些编码器在生成图像检测方面的强能力，其缓解可能有助于改善分布外泛化。

Abstract: Deep vision models exploit shortcuts, relying on cues that correlate with supervision signals. Prior work has focused on visible biases, such as object-background or texture correlations. We identify a different source of shortcut learning: invisible metadata traces embedded at the pixel level, for metadata such as image processing and photo acquisition. We hypothesize that large-scale semantic supervision, whether through categorical labels (ImageNet) or billion-scale captions (LAION), naturally induces metadata-semantics correlations during pretraining, leading models to convert low-level signals into predictive features. By introducing controlled metadata-semantics correlations, we show that stronger ones produce systematically higher sensitivity to metadata traces and larger performance degradation under metadata distribution shifts. We further explore mitigation strategies applied during and after pretraining that reduce sensitivity not only to targeted metadata but also to unseen ones, without sacrificing performance on downstream tasks. Metadata sensitivity also has a positive side: it partly explains the strong generated-image detection ability of some encoders, while its mitigation can improve out-of-distribution generalization. Code: https://github.com/ryan-caesar-ramos/visual-encoder-traces

[36] VideoArgus: Agentic Rubric-Grounded Unified Evaluation for Video Generation and Editing cs.CVPDF

Ziyun Zeng, Zixuan Wang, Yongsheng Yu, Hang Hua, Jiebo Luo

TL;DR: VideoArgus是一个统一的、基于评分标准的框架，用于评估视频生成和编辑任务。它通过为每个输入实例生成一个输出无关的、样本特定的评分标准，并利用该标准指导视觉语言模型和视觉工具进行证据驱动的评分，从而克服了现有基准测试内容固定、覆盖范围有限且评分证据不足的挑战。

Details

Motivation: 现有视频生成评估基准存在依赖固定评估内容、仅覆盖部分生成与编辑场景、且评分证据有限的问题，因此需要一种更统一、更可靠且证据充分的评估框架。

Result: 在一个包含1260个视频的人类对齐数据集上，VideoArgus在五个任务上的斯皮尔曼和肯德尔相关性均高于任务特定的评估器，表明其与人类判断更一致。模型排名在不同评分标准生成和评估VLM骨干网络下也基本保持一致。

Insight: 创新点在于提出了一个统一的、基于可重用评分标准的评估框架，该框架将评估过程分解为具体的标准、评分规则、失败模式和证据计划，并通过视觉语言模型和视觉工具进行证据驱动的自动化评估，提高了评估的透明度和可靠性。

Abstract: Evaluating generated videos remains challenging because existing benchmarks rely on fixed evaluation content, cover only a subset of generation and editing settings, and provide limited evidence for their scores. We introduce VideoArgus, a unified rubric-grounded framework covering five video generation and editing settings. For each input instance, VideoArgus generates an output-blind, sample-specific rubric once and reuses it to evaluate all corresponding candidate videos. The rubric defines concrete criteria, scoring rules, failure modes, and evidence plans, which guide criterion-specific VLM QA and visual tools to produce evidence-grounded criterion scores, rationales, and a diagnostic report. We further construct VideoArgus-Bench, containing 1,026 curated input instances built from 653 high-quality images and 416 high-quality videos, with all benchmark rubrics pre-generated, frozen, and released. On a separate 1,260-video human-alignment set, VideoArgus achieves higher within-input Spearman and Kendall correlations with human judgments than the corresponding benchmark-specific evaluators across all five tasks. Model rankings also remain largely consistent across different rubric-generation and evaluation-VLM backbones. All code and data are released. Visit our project page: https://zzzmyyzeng.github.io/VideoArgus

[37] DynaPix: Can Vision-Language Models Identify the Exact Future? cs.CVPDF

Thong Nguyen, Vinh-Hien Do, Quynh Vo, Cong-Duy Nguyen, See-Kiong Ng

TL;DR: 论文提出了DynaPix基准测试，用于评估视觉语言模型在预测真实未来状态时的能力。该基准要求模型从候选图像中选出视频关键事件后的真实未来图像，场景基于物理模拟器生成，确保真实图像和时间的精确性。研究发现，模型在事件标记目标时刻时表现较好，但在仅靠流逝时间标记时表现接近随机，且人类在后者任务上表现良好，揭示了模型在时间锚定上的差距。

Details

Motivation: 当前评估方法通常接受描述性词语或看似合理的图像，但无法验证预测状态是否与真实未来一致。为了解决这一问题，作者创建了DynaPix基准，旨在使未来预测可检验，从而更准确地评估模型在物理场景中预测真实状态的能力。

Result: 在DynaPix基准上，模型在事件标记目标时刻的任务中表现成功，但在仅靠流逝时间标记的任务中表现接近随机（接近机会水平）。在大型图库搜索任务中，真实图像很少排名第一，表明任务更具挑战性。人类在流逝时间任务上表现良好，而模型则存在困难。通过使用模拟器真实记录进行训练，模型性能得到部分改善，但在较长流逝时间情况下仍不足。

Insight: 论文的创新点在于引入了可检验的未来预测基准DynaPix，通过物理模拟器生成精确的真实未来图像和错误选项，揭示了视觉语言模型在时间锚定上的局限性：模型更擅长将预测锚定到事件上，而非时间本身。这为改进模型的时间理解和物理推理能力提供了方向。

Abstract: Acting in a physical scene requires knowing its real later state, not a plausible one. Current evaluations often accept words or a realistic-looking image, so the predicted state is never checked against the true one. We introduce DynaPix (Dynamic Pixels), a benchmark that makes prediction checkable. Given a video clip that stops before a key event and a question about a later moment, a model must pick the true future image from close candidates or a large gallery. The scenes come from a physics simulator, so the correct image and its time are known exactly and the wrong options are deliberately similar. Models often succeed when a visible event marks the target moment, but are near chance when only elapsed time marks it. Gallery search is harder still, as the true image rarely ranks first. People handle the elapsed-time items well, so the difficulty lies with the models, not the questions. Training on scene accounts drawn from the simulator’s true record, not a teacher’s guess, repairs much of this but not the longer elapsed time case. DynaPix thus exposes a temporal-anchoring gap: models attach a prediction to an event far better than to time itself.

[38] APQF: Agentic Profiling-Guided Structured Pruning and Mixed-Precision Quantization with Adaptive Fine-Tuning cs.CV | cs.AI | cs.LGPDF

Sadegh Jafari, Mohiuddin Bilwal, Fan Zhou, Brian Gelder, Ali Jannesari

TL;DR: APQF是一个自动化框架，旨在通过结构化剪枝和混合精度量化来压缩深度神经网络，以降低其在边缘设备上的计算成本和延迟。该框架利用一个分析代理来评估模型各层的计算成本和对剪枝的敏感性，并基于此指导LLM规划器为每层自动确定剪枝率和量化位宽，同时结合量化感知训练和自适应微调以恢复精度。

Details

Motivation: 现代深度神经网络性能强大但计算成本高，尤其在资源受限的边缘设备上。现有的剪枝和量化方法通常依赖专家手动设置，且采用统一压缩策略，忽略了不同层对压缩的敏感性差异，导致精度损失。

Result: 在ImageNet-1k上，APQF将计算量（比特操作）降至原始模型的5.6-7.7%（即减少13-18倍），同时保持精度接近基线；在20万张图像的预算下，其Top-1准确率比现有联合剪枝量化方法高出约17个百分点。在CIFAR-10上，它在五个架构中的四个上实现了更优压缩。例如，在VGG7上，仅使用基线0.41%的比特操作就达到了93.15%的准确率，且优于其全精度基线。消融实验表明，统一压缩在相同计算量下精度损失最大，且不使用分析数据会损害所有模型性能。

Insight: 创新点在于首次将LLM引导的、基于分析证据的决策与完全训练感知的剪枝和量化流程相结合，实现了针对CNN和视觉Transformer的自动化、分层自适应压缩。其核心是分析代理与LLM规划器的协同，通过数据驱动的层间差异化配置来优化精度与效率的权衡，这为自动化模型压缩提供了新范式。

Abstract: Modern deep neural networks achieve strong performance, but their scale makes them costly and slow, especially on resource-constrained edge devices. Pruning and quantization address this, but rely on manual, expert choices and on algorithms that are hard to apply across architectures. Uniform settings also ignore how differently individual layers respond to compression, which costs accuracy. We introduce APQF, an agentic profiling-guided framework that combines structured pruning, mixed-precision quantization-aware training, and accuracy recovery in one automated pipeline. A profiling agent measures how cost is distributed across the model and how sensitive each part is to pruning, and this evidence drives per-layer pruning ratios, per-layer bit-widths, and the recovery strategy, all proposed by LLM planners and validated before execution. To our knowledge, APQF is the first framework to combine LLM-guided, profiling-grounded decisions with a fully training-aware pruning and quantization pipeline for both CNNs and vision transformers. We evaluate APQF on ResNet, VGG7, ViT, DeiT, and Swin using ImageNet-1k and CIFAR-10. On ImageNet it cuts compute to 5.6-7.7 percent of the original bit-operations, a 13-18x reduction, while keeping accuracy close to the baseline, and under a 200K-image budget it stays roughly 17 points higher in Top-1 than existing joint pruning and quantization methods. On CIFAR-10 it compresses further than that method on four of five architectures. On VGG7 it reaches 93.15 percent using only 0.41 percent of baseline bit-operations, the only method at that compression level to improve on its full-precision baseline. Ablations show that uniform compression loses the most accuracy at matched compute, and that withholding profiling data from the planner hurts every model. Six LLM planners, including free open-weight ones, all reach 97.4-97.9 percent on Swin-Tiny.

[39] HERA: Historical Evidence Routing Adapter for Physical Prediction in Latent World Models cs.CVPDF

Yuanruyi, Yue Cao, Haojia Gao, Guanqiu Guo, Ziyuezhang

TL;DR: 论文提出了HERA（历史证据路由适配器）框架，用于解决潜在世界模型中遮挡场景下的物理预测难题。该框架通过引入Register-Routed Patch Memory（RRPM）轻量级适配器，将保留的历史证据选择性地路由到冻结的潜在预测器中，而无需干扰其原有的潜在工作空间。

Details

Motivation: 现有预测视频模型在处理遮挡物理事件时面临挑战，因为后续预测可能依赖于当前视图中已不可用的物体证据。现有方法主要通过扩大时间上下文、缓存通用视频特征或强加显式的以物体为中心的状态来改进，但未能直接解决如何在不干扰预训练预测器原生潜在工作空间的情况下，选择性地检索和整合相关历史证据的问题。

Result: 在IntPhys2 Main数据集上，HERA与RRPM将V-JEPA 2-G模型的成对AvgSurprise准确率从52.57%提升至54.35%。子组分析显示，在固定摄像头连续性任务上从46.15%提升至57.69%，在固定摄像头不变性任务上从46.15%提升至63.46%。

Insight: 创新点在于提出了一个轻量级适配器框架，通过结构化记忆库、记忆寄存器和工作室寄存器，实现了对历史证据的选择性路由和整合，这是一种实用的适应策略，可以在不重新训练或干扰原有预测器的情况下，有效增强模型对遮挡物理事件的预测能力。

Abstract: Predictive video models have emerged as promising world models by learning latent visual dynamics from large-scale video. Yet these models remain challenged by physical events under occlusion, where later predictions may depend on object evidence that is no longer available in the current view. Addressing this challenge requires historical evidence not only to be preserved but also to remain accessible when it becomes relevant to a subsequent prediction. Existing approaches mainly enlarge the temporal context, cache generic video features, or impose explicit object-centric states, thereby improving the capacity or structure of retained history. However, they do not directly address how relevant historical evidence can be selectively retrieved and integrated into a pretrained predictor without interfering with its native latent workspace. Accordingly, we introduce HERA (Historical Evidence Routing Adapter), a framework for routing retained historical evidence into a frozen latent predictor, and instantiate it with Register-Routed Patch Memory (RRPM), a lightweight adapter comprising a Structured Memory Bank, Memory Registers, and Workspace Registers. On the IntPhys2 Main split, HERA with RRPM improves the pairwise AvgSurprise accuracy of V-JEPA 2-G from 52.57% to 54.35%. Subgroup analysis shows particularly strong improvements on fixed-camera continuity, from 46.15% to 57.69%, and fixed-camera immutability, from 46.15% to 63.46%. These results support historical evidence routing as a practical adaptation strategy for physical prediction in latent world models.

[40] OmniMech: All-in-one Multimodal Mechanical Benchmark for 3D Reconstruction cs.CVPDF

Taiting Lu, Runze Liu, Ziwei Dong, Sisong Bei, Jingying Zeng

TL;DR: 本文提出了OmniMech，首个面向工业制造数据、用于评估视觉语言模型生成可执行CAD程序能力的百万级基准。该基准包含超过25.1万张带完整尺寸和公差标注的2D正交工程图，并配对了原生CAD模型、多视图渲染、网格等多种表示形式以及丰富的语义标注。它定义了从工程图合成参数化CAD程序、基于图纸的3D推理、基于标注的推理以及工具增强的智能体推理四个核心任务。

Details

Motivation: 当前视觉语言模型主要针对通用、粗略的3D物体生成，而工业机械设计需要细粒度几何和毫米级公差，现有方法难以满足。因此，需要建立一个专门的基准来评估和推动模型在工业制造场景下的能力。

Result: 实验表明，当前的视觉语言模型和CAD专用模型在可执行程序合成、细粒度3D重建以及可靠执行尺寸和公差约束方面仍然存在困难。

Insight: 创新点在于构建了首个大规模、多任务、多表示的工业机械设计基准，并强调了将制造约束（如尺寸公差）和工具增强推理整合到CAD生成评估中的重要性，为面向精确工程的设计AI研究提供了关键基础设施。

Abstract: Recent vision-language models (VLMs) can generate executable CAD programs from images, but existing methods mainly target coarse, general-purpose 3D objects and rarely address the fine-grained geometry and millimeter-level tolerances required in industrial mechanical design. We introduce OmniMech, the first million-scale benchmark for evaluating VLMs on executable CAD generation from industrial manufacturing data. OmniMech contains more than 251,000 fully dimensioned and toleranced 2D orthographic drawings, paired with native CAD models, multi-view renderings, mesh, STEP and B-rep representations, and rich semantic annotations. The benchmark includes four tasks: (1) parametric CAD program synthesis from engineering drawings; (2) diagram-to-3D reasoning for geometrically and structurally consistent reconstruction; (3) annotation-grounded reasoning over dimensions, symbols, feature callouts, and manufacturing constraints; and (4) tool-augmented agentic reasoning using visualization, measurement, CAD execution, and verification tools. Experiments show that current VLMs and CAD-specialized models still struggle with executable program synthesis, fine-grained 3D reconstruction, and reliable enforcement of dimensions and tolerances. We will release the benchmark data, evaluation code, and tool interfaces to support future research.

[41] From Sports to Safety: Benchmarking Proactive Risk Inference in MLLMs cs.CV | cs.CLPDF

Jiawei Qiu, Yichen Xu, Jianzhe Ma, Mingyang Yu, Wenbin Zhu

TL;DR: 该论文提出了SPRINT基准测试，用于评估多模态大语言模型在体育视频中主动预测物理风险的能力。该基准包含2888个真实世界体育视频，涵盖14种运动和3种环境设置，并标注了早期危险线索、事故时间和分层原因。评估发现，现有MLLMs在危险信号检测上表现良好（>95%），但在识别原因上表现不佳（<50%），且容易在安全视频上产生误报。

Details

Motivation: 现有MLLM评估主要关注有害内容或一般风险，而主动预测物理危险（对现实世界安全至关重要）尚未得到充分探索。体育运动因其伤害原因多样且事故前时空线索需要推理能力，成为研究更广泛安全领域（如自动驾驶、跌倒检测）的理想测试平台。

Result: 在SPRINT基准上评估最先进的MLLMs发现，最佳模型在发出危险信号方面超过95%，但在识别事故原因方面低于50%。诊断实验进一步表明，即使在没有危险的视频上，明确的危险查询也会引发严重的误报。

Insight: 论文的创新点在于构建了首个专注于主动物理风险推理的体育视频基准（SPRINT），并揭示了当前MLLMs仅具备表面的主动安全能力，缺乏稳定、基于原因的预警。这强调了在动态物理环境中开发可靠主动安全系统的必要性。

Abstract: Timely anticipation of physical hazards is essential for real-world safety, yet existing MLLM evaluations focus on harmful content or general risks, leaving proactive physical hazard prediction underexplored. Sports provide a well-suited testbed: accident causes span diverse injury dimensions and pre-accident spatiotemporal cues draw on reasoning capabilities shared with broader safety domains such as autonomous driving and fall detection. We introduce SPRINT (Sports Proactive Risk INference Testbed), a benchmark of 2,888 real-world sports videos (2,440 accident, 448 safe controls) spanning 14 sports and 3 environmental settings. Accident videos feature fine-grained annotations of early hazard cues, accident timing, and hierarchical causes; safe videos are manually verified as accident-free and serve to diagnose prompt-induced false alarms. Evaluating state-of-the-art MLLMs under diverse prompts and temporal windows reveals a sharp gap between hazard sensitivity and understanding: the best model exceeds 95% in signaling hazards yet falls below 50% in identifying their causes. Diagnostic experiments further show that explicit danger queries trigger severe false alarms even on hazard-free videos. These findings indicate that current MLLMs exhibit only superficial proactive safety, lacking stable, cause-grounded early warning, and underscore the need for reliable proactive safety in dynamic physical environments. Data and code will be open-sourced upon acceptance.

[42] EffectLearner: World-Aware Object-Effect Reasoning for Real-World Video Object Removal cs.CVPDF

Feier Wu, Wanke Xia, Xu He, Zilang Zhou, Si Chen

TL;DR: EffectLearner是一个用于真实世界视频物体移除的框架，它通过结合基于视觉语言模型（VLM）的物体-效应推理器和基于扩散Transformer（DiT）的视频擦除器，不仅移除目标物体，还移除其引发的各种复杂效应。该框架利用结构化的效应分析提示进行跨模态推理，并辅以运动感知的掩码引导和一致性监督，以提升移除的覆盖范围和时空稳定性。

Details

Motivation: 现有视频物体移除方法主要从预定义的效应类别和固定数据分布中隐式学习物体-效应对应关系，难以泛化到包含组合效应、空间分离或弱相关效应、长尾物理现象以及动态交互的复杂真实世界场景。

Result: 在标准基准ROSE-Bench上，EffectLearner在多数指标上超越了现有基线方法；在专门构建的复杂效应数据集EffectWorld的评估集（EffectWorld-Eval）和更具挑战性的野外集（EffectWorld-Wild）上，也展现出明显优势，证明了其在复杂真实场景中实现高质量视频物体移除的能力。

Insight: 核心创新在于提出了一个显式的、基于语义推理的物体-效应分析模块（Object-Effect Reasoner），将效应感知的上下文信息结构化地注入到视频修复过程中。此外，构建专门针对复杂物体诱发效应的数据集EffectWorld并采用渐进式训练课程，有效提升了模型对真实世界复杂物理交互的泛化能力。

Abstract: Video object removal must eliminate not only the target object but also its induced effects while maintaining high-fidelity and spatiotemporally coherent restoration. Existing methods mainly learn object-effect correspondences implicitly from predefined effect categories and fixed data distributions, limiting their generalization to complex real-world scenes involving compositional effects, spatially detached or weakly correlated effects, long-tail physical phenomena, and dynamically evolving interactions. We propose EffectLearner, a semantic-reasoning-enhanced framework that combines a VLM-based Object-Effect Reasoner with a DiT-based Video Eraser. Guided by a structured effect-analysis prompt, the Reasoner performs cross-modal reasoning over a target-highlighted video and extracts compact effect-aware context, which guides the Video Eraser toward comprehensive object-effect removal. Motion-aware mask guidance and motion-consistency supervision further improve removal coverage and spatiotemporal stability under object motion and evolving scene dynamics. To fully exploit the framework in challenging real-world scenarios, we further construct EffectWorld, a paired video dataset specifically designed for complex object-induced effects, and introduce a progressive training curriculum that combines common supervision with complex-effect data. On the standard ROSE-Bench, EffectLearner outperforms existing baselines on most metrics and achieves clear advantages on both EffectWorld-Eval and the challenging EffectWorld-Wild, demonstrating its ability to deliver high-quality video object removal in complex real-world scenes.

[43] Beyond Frame Selection: Rethinking Long-Video Understanding with MLLMs cs.CVPDF

Ziling Huang, Shin’ichi Satoh

TL;DR: 本文提出VideoRouter(VR)方法，重新思考长视频理解问题，将其视为协调互补证据视图而非选择单一帧子集。该方法通过构建与问题无关的时间层次结构，将视频划分为从粗到细的时间连贯片段，形成全局视图和局部视图，并引入验证引导的路由器来选择最终答案。

Details

Motivation: 现有方法通常依赖均匀采样或帧选择策略，难以同时保留全局故事情节上下文和细粒度证据，导致长视频理解面临挑战。

Result: 在VideoMME基准测试中，使用LLaVA-Video-7B骨干网络，该方法比最先进的帧选择方法性能提升2.9个百分点。

Insight: 创新点在于将长视频理解重新定义为协调互补证据视图，通过时间层次结构自然形成全局和局部视图，并引入验证引导的路由器进行协调，而非传统的帧选择策略。

Abstract: Multimodal Large Language Models (MLLMs) have achieved strong progress in video understanding, yet it remains challenging because the token limitation makes MLLMs difficult to capture temporally sparse evidence. Existing methods typically rely on uniform sampling, or frame selection, but these strategies usually optimize either broad temporal coverage or local relevance, making it difficult to preserve both global storyline context and fine-grained evidence. We propose VideoRouter(VR) that rethinks long-video understanding as coordinating complementary evidence views rather than selecting a single subset of frames. It first organizes each video into a question-agnostic temporal hierarchy, which partition the video into coarse-to-fine temporally coherent segments. In this hierarchy, upper-level nodes capture broad storyline context and event progression, while lower-level nodes preserve fine-grained local details and evidence-bearing moments. This naturally gives rise to two complementary views: a global view for coverage-oriented reasoning and a local view for detail-oriented evidence recovery. We further introduce a verification-guided router to determine which view is better supported by the selected evidence and select the final answer. We validate the effectiveness of the proposed design through extensive experiments, showing that the verification-guided router effectively coordinates global and local reasoning, and that, on VideoMME, our method outperforms state-of-the-art frame selection methods by 2.9 points, respectively, under the LLaVA-Video-7B backbone. We will release the code.

[44] TruthLens: Object Hallucination Detection via Self-Evaluating Truthfulness Scores in LVLMs cs.CVPDF

Yanqi Wu, Runhe Lai, Xinhua Lu, Qichao Chen, Zhiping Zhou

TL;DR: 本文提出TruthLens，一种用于检测大型视觉语言模型（LVLM）中物体幻觉的自评估框架。其核心是利用一个特殊标记作为参考，从语言建模头中提取每个物体标记的‘真实性分数’，并通过微调使该分数能区分真实与幻觉物体。该方法无需额外模型或推理成本，且在多个LVLM上实现了最先进的检测性能。

Details

Motivation: 尽管LVLM取得了显著进展，但物体幻觉问题阻碍了其可信部署。研究发现，真实与幻觉的物体标记在隐藏表示中是可分离的，但这种可分离性在语言建模头中基本丢失，因此需要一种方法让LM头能暴露物体真实性信号。

Result: 在多个LVLM上的广泛实验证明了其最先进的性能。特别是在Qwen2.5-VL-7B模型上，TruthLens在MS-COCO基准上的AUROC指标比之前的最佳方法高出超过17%。

Insight: 创新点在于重新利用一个特殊标记作为参考，从LM头中直接提取可解释的‘真实性分数’，并通过一个结合MSE损失和发散约束的微调目标，使模型在保持原有生成能力的同时，获得了强大的、可泛化的幻觉检测能力，且无需额外推理开销。

Abstract: Despite the remarkable progress of large vision language models (LVLMs), object hallucination remains a fundamental challenge that hinders their trustworthy deployment. A key finding motivates our work: real and hallucinated object tokens are clearly separable in hidden representations, yet this separability is largely lost at the language-modeling (LM) head. We propose TruthLens, a self-evaluation framework that teaches the LM head to expose a per-object truthfulness signal without any auxiliary model or additional inference cost. Concretely, a rarely-used special token is repurposed as a reference token. For each object-token position, we extract the log-probability assigned to this special token by the LM head, and define its difference from a predefined constant as the truthfulness score. The model is then fine-tuned with an MSE objective that drives scores toward 1 for real objects and 0 for hallucinated ones, while a divergence constraint preserves the original generation capability. Despite being trained on only a limited set of object categories, TruthLens generalizes effectively to benchmarks with substantially larger label spaces. Extensive experiments across multiple LVLMs demonstrate state-of-the-art performance; notably, on Qwen2.5-VL-7B, TruthLens outperforms the previous best method on MS-COCO by over 17% in AUROC. Our code is available at https://github.com/wyqstan/TruthLens.

[45] SCI-CLIP: Segment-Centric Inference with Reference Memory for Training-Free Open-Vocabulary Segmentation cs.CVPDF

Mohamad Zamini, Diksha Shukla

TL;DR: SCI-CLIP是一个无需训练的开集词汇分割框架，它围绕区域抽象原则构建，通过诱导区域一致的交互图、传播特征值、选择性跨窗口支持以及查询离线参考记忆，将冻结的CLIP风格特征转化为空间连贯、上下文感知且兼容检索的密集预测。

Details

Motivation: 当前无需训练的开集词汇分割方法受限于缺乏合适的推理抽象，因为冻结的视觉语言特征在图像块级别产生，而密集预测需要一个能同时管理特征交互、空间支持、上下文恢复和基于检索的校正的单元。

Result: SCI-CLIP在八个基准测试中一致提升了密集预测的结构质量、上下文推理的鲁棒性以及基于示例校正的对齐性，从而实现了更强的开集词汇分割性能。

Insight: 创新点在于提出了以区域为中心的推理框架，使用相同的区域抽象来组织密集开集词汇预测的所有阶段，包括特征交互、上下文恢复和检索对齐，这避免了训练需求并提升了预测的连贯性和兼容性。

Abstract: Training-free open-vocabulary segmentation remains limited by a missing inference abstraction. Frozen vision-language features are produced at patch level, yet dense prediction requires a unit that simultaneously governs feature interaction, spatial support, contextual recovery, and retrieval-based correction. We present SCI-CLIP, a segment-centric inference framework built around the principle that the same region abstraction should organize all stages of dense open-vocabulary prediction. SCI-CLIP first induces a region-consistent interaction graph over frozen visual tokens, then reconstructs dense features by propagating values over this graph, augmenting them with selective cross-window support only where local evidence is insufficient. The same segment abstraction is subsequently used to construct and query an offline reference memory, aligning exemplar retrieval with the units on which prediction is made. SCI-CLIP turns frozen CLIP-style features into spatially coherent, context-aware, and retrieval-compatible dense predictions without any training. SCI-CLIP consistently improves the structural quality of dense predictions, the robustness of contextual reasoning, and the alignment of exemplar-based correction, yielding stronger open-vocabulary segmentation across eight benchmarks. Project code is available at: https://github.com/mzamini92/SCICLIP.

[46] ChronoVision: Temporal Reasoning via Latent State Reconstruction cs.CVPDF

Yifan Shen, Jian Xu, Boyi Li, Yuner Zhang, Tianjiao Yu

TL;DR: 本文提出了ChronoVision，一个旨在提升多模态大语言模型时序推理能力的框架。它通过重构最终状态的潜在表示和对关键视觉证据的关注来对齐视觉逻辑，并采用强化学习进行后训练。此外，论文还引入了Vbvr-VQA数据集来严格评估时序跟踪能力。

Details

Motivation: 多模态大语言模型在被动感知方面表现出色，但在需要多步时序推理的复杂视觉认知任务上存在困难，这主要源于基于语言的推理固有的模糊性，难以准确描述连续的视觉变换。

Result: 实验表明，ChronoVision在Vbvr-VQA数据集上取得了SOTA性能，域内和域外准确率分别为74.8%和71.6%，同时在极具挑战性的跨域基准IntPhys2上也达到了55.0%的准确率。

Insight: 创新点包括：1）通过重构最终状态的潜在表示来显式地对齐视觉逻辑；2）利用ROI注意力定位模块和语义跨度查询聚焦关键视觉证据；3）采用结合结果正确性、潜在过程对齐和无监督视觉关注的复合奖励函数进行强化学习后训练；4）提出了将视频推理重构为严格图像排序任务的Vbvr-VQA评估数据集。

Abstract: Multimodal large language models excel at passive perception but struggle with complex visual cognitive tasks requiring multi-step temporal reasoning. This degradation largely stems from the inherent ambiguity of language-based reasoning, which often fails to accurately articulate continuous visual transformations. To address this, we propose ChronoVision, a multimodal framework designed to align visual logic with latent imagery. During supervised fine-tuning, a Reconstructive Visual Head predicts the latent representation of the final transformed state, while an ROI Attention Locating module focuses the model on key visual evidence via semantic span queries. In post-training, we apply reinforcement learning with an implicit process grounding mechanism, guided by a composite reward function that evaluates outcome correctness, latent process alignment, and unsupervised visual focus. Furthermore, we introduce Vbvr-VQA, a novel dataset that evaluates temporal tracking by reformulating video reasoning into a strict image-ordering task. Experiments demonstrate that ChronoVision achieves state-of-the-art performance on Vbvr-VQA with 74.8% in-domain and 71.6% out-of-domain accuracy, alongside a strong 55.0% accuracy on IntPhys2, a highly challenging cross-domain benchmark.

[47] Vorch-Streamer: Extending Human Audio-Visual Generation to Real-Time Long-Form Streaming cs.CV | cs.SDPDF

Menglin Han, Yang Ding, Yulei Lu, Haoran Yu, Xin Ma

TL;DR: Vorch-Streamer是一个后训练框架，旨在解决预训练双向模型在实时长时音频-视频生成中面临的因果性、连续合成挑战，实现了从文本到音视频的实时流式生成。

Details

Motivation: 动机是解决将预训练双向模型适配到实时长时流式生成场景时的两个关键困境：自回归重用生成块作为上下文导致的曝光偏差和视觉漂移，以及全局语音话语无法在有限局部上下文下指导因果生成器决定下一段应说的内容。

Result: 在合成的8万条12-21秒虚拟人片段数据集上训练，模型以27.12 FPS联合生成音频和视频，超过24 FPS的实时播放速率，并在长时生成中保持了有竞争力的音频-唇部同步和强身份保持能力。

Insight: 创新点包括：采用混合教师强制和扩散强化的因果生成器训练，结合长时域自强制与DMD蒸馏以保持预训练教师质量；引入外部语言模型预测离散语音规划标记，显式控制语音进程并确保因果块与内容对齐。

Abstract: Real-time long-form avatar audio–video generation requires causal, continuous synthesis while maintaining audiovisual synchronization and visual consistency. Adapting a pretrained bidirectional model to this setting presents two key dilemmas. First, autoregressively reusing generated blocks as context creates exposure bias, causing errors and visual drift to accumulate over long rollouts. Second, a global speech utterance does not indicates a causal generator which portion should be spoken next when only limited local audio–video context is available. We present \textbf{Vorch-Streamer}, a post-training framework that addresses these challenges and enables real-time long-form Text-to-Audio-Video (T2AV) streaming. We construct a synthetic corpus of 80K avatar clips spanning 12–21 seconds and first train a causal generator with mixed Teacher Forcing and Diffusion Forcing. We then apply long-horizon Self Forcing with DMD distillation, exposing the model to its own rollout distribution while preserving the quality of the pretrained bidirectional teacher. To explicitly control speech progression, an external language model predicts discrete 25-Hz speech-planning tokens, whose continuous features condition the audio diffusion branch and align each causal block with the content it should speak. With bounded causal context and four-step denoising, Vorch-Streamer jointly generates audio and video from text at 27.12 FPS, exceeding the 24-FPS real-time playback rate while maintaining competitive audio–lip synchronization and strong identity preservation over long-form generation.

[48] Vorch-IR: Long-Form Unified Multimodal Identity Replacement Video Generation cs.CVPDF

Yaole Wang, Xiaoyu Chen, Xin Ma, Yang Ding, Gang Yue

TL;DR: 本文提出了Vorch-IR，一个统一的视频身份替换框架，支持单人和双人身份替换，并可选择性地替换背景。该方法基于LTX2架构，通过联合条件化驱动视频、索引参考图像和文本编辑指令来工作，无需特定任务的结构控制。作者还开发了一个自动数据构建流程来生成配对监督数据，并通过实验验证了其在身份保持、运动保真度和时间一致性方面的优越性能。

Details

Motivation: 现有视频身份替换方法主要针对单人场景，且通常需要任务特定的结构控制（如掩码或姿态表示），限制了其在通用多模态编辑系统中的灵活性。此外，多人替换的进展因缺乏配对训练数据而受到制约。

Result: 实验使用自动指标和成对人类评估表明，该方法在多样化场景中实现了强大的身份保持、运动保真度和时间一致性。通过时间重叠推理策略，该短片段模型还能扩展到生成长达数分钟的视频，而无需自回归延续。

Insight: 主要创新点在于提出了一个统一的、指令驱动的框架，通过文本指令指定参考图像的角色（主体或背景），并利用自注意力和跨注意力融合密集视觉条件与视觉-语言上下文以建立语义对应。此外，自动数据构建流程解决了多人替换任务中配对数据稀缺的问题。

Abstract: Video identity replacement seeks to transfer the identities of one or more subjects while preserving the motion, expressions, and temporal structure of a driving video. Existing methods largely target single-person settings and often require task-specific structural controls, such as masks or pose representations, limiting their flexibility in general multimodal editing systems. Progress on multi-person replacement is further constrained by the scarcity of paired training data. We present Vorch-IR, a unified framework that supports single- and dual-person identity replacement, with optional background replacement, in a single model. Built on LTX2, Vorch-IR jointly conditions on a driving video, indexed reference images, and a textual editing instruction. The reference images need not match the pose, layout, or spatial configuration of the driving video: their roles as subject or background references are specified through the instruction. Dense visual conditions are fused through self-attention, while a vision-language context establishes semantic correspondence through cross-attention. We further develop an automatic data construction pipeline that synthesizes paired supervision for all four editing settings. Experiments using automatic metrics and pairwise human evaluation demonstrate strong identity preservation, motion fidelity, and temporal coherence across diverse scenarios. A temporal overlapping inference strategy additionally extends the short-clip model to minute-long generation without autoregressive continuation.

[49] DistMedVL: Distributional Vision-Language Alignment for Uncertainty-Aware Medical Image Segmentation cs.CV | cs.AI | cs.MMPDF

Jiaxuan Li, Qing Xu, Xiangjian He, Yue Li, Daokun Zhang

TL;DR: 本文提出了DistMedVL，一个用于不确定性感知医学图像分割的概率视觉-语言对齐框架。该方法通过一个轻量级的概率跨模态适配器（PCM-Adapter），在冻结的编码器之上显式建模表示不确定性，以解决真实临床条件下多模态不确定性对跨模态对齐的阻碍。

Details

Motivation: 现有视觉-语言分割方法依赖于确定性的跨模态匹配，忽略了来自模糊边界的偶然性不确定性和有限训练数据带来的认知不确定性，导致在领域偏移下性能脆弱。本文旨在解决这一问题。

Result: 在八个医学分割基准上的广泛实验表明，DistMedVL仅用630万可训练参数就超越了最先进的方法，表现出卓越的数据效率、扰动鲁棒性和跨数据集泛化能力。

Insight: 核心创新点在于将文本标记建模为高斯分布，并使用马氏距离计算补丁-文本兼容性，实现方差加权的匹配；同时设计了分布流模块，通过视觉引导精炼文本分布，以适应不同成像模态间的分布变化。这是一种显式建模不确定性的概率对齐方法。

Abstract: Cross-modal alignment of visual and textual representations is fundamental to multimodal medical image understanding, yet remains hindered by uncertainty in both modalities under real-world clinical conditions. Existing vision-language segmentation methods rely on deterministic cross-modal matching, which overlooks aleatoric uncertainty from ambiguous boundaries and epistemic uncertainty from limited training data, leading to fragile performance under domain shift. To address this issue, we propose DistMedVL, a probabilistic vision-language framework that introduces a lightweight Probabilistic Cross-Modal Adapter (PCM-Adapter) upon frozen encoders to explicitly model representational uncertainty. Specifically, the PCM-Adapter comprises two sequential modules for progressive probabilistic alignment. We first devise a Mahalanobis Alignment Module (MAM) that models textual tokens as Gaussian distributions and computes patch-text compatibility via Mahalanobis distance, yielding variance-conditioned matching that downweights unreliable feature dimensions. Moreover, we devise a Distribution Flow Module (DFM) that estimates modality-wise confidence parameters and performs vision-guided refinement of textual distributions, accommodating distributional variation across imaging modalities. Extensive experiments across eight medical segmentation benchmarks demonstrate that DistMedVL outperforms state-of-the-art methods with only 6.3M trainable parameters, exhibiting superior data efficiency, perturbation robustness and cross-dataset generalization.

[50] URNet: A Unified Reparameterized Network for Efficient RGB-D Semantic Segmentation cs.CVPDF

Guoan Xu, Zhengxue Wang, Yang Xiao, Ligeng Chen, Guangwei Gao

TL;DR: 本文提出URNet，一种统一重参数化的RGB-D语义分割网络，通过单一编码器同时执行多模态特征提取和跨模态融合，并引入线性门控注意力模块和金字塔合并解码器，在多个基准测试中实现了最先进的性能和高效率。

Details

Motivation: 现有RGB-D语义分割方法通常采用双编码器分别处理RGB和深度输入，导致深度表示捕获不足、跨模态交互受限，且额外编码器分支引入冗余计算，阻碍轻量化执行。

Result: 在多个RGB-D分割基准测试上进行广泛实验，URNet实现了最先进的性能，同时保持了高效率。

Insight: 创新点包括采用重参数化策略压缩网络架构以促进快速推理，在重参数化块中引入线性门控注意力模块以充分利用不同特征尺度的互补RGB和深度线索，并设计了一个简洁有效的通用解码器（金字塔合并解码器），弥补了现有模型在解码器设计上的不足。

Abstract: Previous RGB-D semantic segmentation methods commonly employ dual encoders to separately process RGB and depth inputs, followed by dedicated modules for cross-modal feature fusion. However, such designs often inadequately capture depth representations and consequently limit effective cross-modal interaction, while the additional encoder branch introduces redundant computation that hinders lightweight execution. To tackle these challenges, we propose URNet, a Unified Reparameterized RGB-D Network that performs simultaneous multi-modal feature extraction and cross-modal fusion within a single encoder. Specifically, we adopt a reparameterization strategy to compact the network architecture and facilitate fast inference. Within each Reparameterized Block (RepBlock), a Linear Gated Attention (LGA) module is introduced to fully exploit complementary RGB and depth cues across different feature scales. Furthermore, considering that decoder design has been relatively underexplored in existing RGB-D segmentation models, we develop a concise yet effective universal decoder, termed the Pyramid Merging Decoder (PMD). Extensive experiments on multiple RGB-D segmentation benchmarks demonstrate that URNet achieves state-of-the-art performance while maintaining high efficiency. Code will be available at https://github.com/Wild-Stephen/URNet.

[51] SciQNet: Two-Stage Multimodal Adaptation for Scientific Image Quality Assessment cs.CVPDF

Yin-Loon Khor, Yi-Jie Wong, Jing Jie Tan, Ming Jie Lee

TL;DR: 本文提出了SciQNet，一个用于科学图像质量评估的两阶段多模态适应框架。该框架首先在科学文档图像上进行领域自适应预训练，然后通过联合评分和理解监督进行任务特定的微调。实验表明，使用40%的分层子集进行预训练效果最佳，最终模型在ICME 2026挑战赛的评分赛道中排名第二。

Details

Motivation: 科学图像的质量评估不仅依赖于视觉清晰度，还涉及科学信息性，这比自然图像更具挑战性，因此需要专门的方法来同时评估这两个方面。

Result: 模型在SIQA-S评分上达到92.21，SIQA-U评分上达到47.38，综合得分为69.80，在ICME 2026科学图像质量评估挑战赛的评分赛道中排名第二。

Insight: 创新点包括两阶段适应框架（领域自适应预训练和任务特定微调）以及联合监督策略（结合指令调优和Huber损失进行评分监督，以及多选视觉问答进行理解监督），并发现预训练数据相关性可能比规模更重要。

Abstract: Scientific images are essential for communicating experimental observations, quantitative evidence and conceptual knowledge. Unlike natural images, their quality depends on both visual clarity and scientific informativeness, making assessment challenging. In this work, we present SciQNet, a two-stage multimodal adaptation framework for scientific image quality assessment. The first stage performs domain-adaptive pretraining on scientific document images and the second stage conducts task-specific fine-tuning with joint scoring and understanding supervision. For scoring-oriented supervision, we combine instruction tuning with a Huber loss derived from rating-word logits, while understanding-oriented supervision is formulated as multiple-choice visual question answering. Experiments show that using a 40% stratified subset of the domain-adaptive data gives the best performance among the evaluated pretraining fractions, suggesting that pretraining-data relevance may be as important as pretraining-data scale. The final model achieves an SIQA-S score of 92.21, an SIQA-U score of 47.38 and a combined score of 69.80. This work presents our solution to the ICME 2026 Scientific Image Quality Assessment Challenge, which ranked 2nd in the scoring track.

[52] TAU-Bench: From Anomaly Instance Tracking to Fine-Grained Video Anomaly Understanding cs.CVPDF

Kepeng Yang, Dongxuan Liu, Rongxin Gao, Zixin Su, Rui Wu

TL;DR: 本文提出了TAU-Bench，一个以轨迹为中心的基准测试，用于联合评估视频异常实例跟踪和细粒度异常理解能力。该基准包含大量视频、轨迹和像素级掩码，并集成了实例识别、事件理解和场景推理的标注。评估发现，现有视觉语言模型在生成合理异常描述时，仍难以可靠地定位和跟踪正确的异常实例，揭示了语义推理与视觉基础之间的差距。

Details

Motivation: 现有基准通常将跟踪和语义理解分开评估，导致模型在生成流畅的异常描述时，可能无法在时间上持续地基于正确的异常实例进行解释，即存在实例-语义不一致的问题。

Result: 在TAU-Bench上对代表性视觉语言模型家族的评估表明，模型即使能生成合理的异常解释，仍可能无法可靠地定位和跟踪正确的实例，这凸显了语义推理与视觉基础之间的持续差距。

Insight: 论文的创新点在于提出了一个联合评估异常实例跟踪与细粒度异常理解的基准，并揭示了当前模型在实例层面的视觉基础能力不足，强调了面向实例的评估对于实现更忠实可靠的视频异常理解系统的重要性。

Abstract: Humans understand anomalous events through a coherent perceptual process in which they identify the focal instance, follow its behavior as the event unfolds, and interpret why it violates the expectations of the surrounding scene. Video anomaly understanding (VAU) seeks to endow models with a similar capability, moving beyond deciding whether a video is anomalous toward explaining how the event develops and why it matters. Although recent vision–language models (VLMs) can generate detailed and plausible anomaly descriptions, their semantic fluency does not ensure that these interpretations remain grounded in the correct anomaly instance over time. Existing benchmarks typically evaluate tracking and semantic understanding through separate protocols, leaving such instance–semantic inconsistency largely unmeasured. We therefore introduce TAU-Bench, a track-centric benchmark for jointly evaluating anomaly instance tracking and fine-grained anomaly understanding. TAU-Bench contains 1,118 videos, 1,454 tracks, and 202,438 pixel-level masks spanning 49 event and 45 scene categories, together with track-centric annotations that connect instance-level identification, event-level understanding, and scene-level reasoning. To build TAU-Bench at scale, we developed an automated data engine integrating anomaly suitability filtering, anomaly instance track construction, hierarchical caption annotation, and human quality control. Evaluations across representative VLM families show that models producing plausible anomaly interpretations may still fail to localize and track the correct instance reliably, revealing a persistent gap between semantic reasoning and visual grounding. These findings therefore highlight instance-grounded evaluation as an important step toward more faithful and reliable VAU systems.

[53] StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding cs.CVPDF

Xichen Zhang, Guankai Li, Yinghao Zhu, Shijian Wang, Sitong Wu

TL;DR: 论文提出了StreamArena基准测试和StreamMind架构，用于解决智能体在连续、长时间、交互式视频流理解中的挑战。StreamArena包含长视频和开放式问答对，评估实时感知、历史回顾、主动交互和多模态工具使用能力。StreamMind采用两层架构，通过前端工作者处理低延迟交互，后端工作者异步构建持久化多模态记忆，从而在性能和延迟上超越现有基线。

Details

Motivation: 当前自主多模态智能体的评估主要依赖短视频片段和选择题格式，这导致仅处理最近几帧的简单基线模型就能达到甚至超越复杂流式模型的性能，且选择题选项会暴露语言捷径。因此，需要一个新的基准来评估智能体在小时尺度、交互式流媒体视频中的真实理解能力。

Result: 在提出的StreamArena基准测试上，StreamMind架构在实时感知、历史回顾、主动交互和多模态工具使用这四项能力上均优于现有的流式基线模型，并通过复用持久化状态降低了查询到回答的延迟。

Insight: 论文的核心创新点在于揭示了连续交互与长时程多模态理解之间的内在矛盾，并提出了一个解耦的两层架构（StreamMind）来解决它。该架构将低延迟的交互任务与需要长期、精细记忆的异步处理任务分离，有效平衡了性能与效率，为构建实用的长时程流媒体智能体提供了新思路。

Abstract: Deploying autonomous multimodal agents in continuous, real-world environments requires them to ingest unbounded audio-visual streams and maintain hour-scale memory. However, current evaluations predominantly rely on brief clips and multiple-choice formats. This design allows minimal baselines that process only the last four frames to match or surpass complex streaming models, while answer options also expose language shortcuts. We introduce StreamArena, a benchmark for hour-scale, interactive streaming video understanding. StreamArena contains 243 full-length videos averaging 88.8 minutes and 3,646 rigorously annotated, open-ended question-answer pairs that evaluate real-time perception, historical retrospection, proactive interaction, and multimodal tool utilization. Evaluation across diverse systems exposes a tension between continuous interaction and long-horizon multimodal comprehension. Methods that retain only recent frames cannot recover distant events, methods that convert past observations into text lose visual evidence, and methods that repeatedly compress visual memory struggle to preserve fine-grained details over time. We address this tension with StreamMind, a two-tier architecture that assigns latency-critical interaction and proactive monitoring to independently scheduled frontend workers, while backend workers asynchronously construct persistent multimodal memory and perform historical recall and external search. StreamMind outperforms existing streaming baselines across all four capabilities and reduces query-to-answer latency by reusing persistent state.

[54] G$^2$ARD-GS: Geometry-Guided Anchor-Regularized Gaussian Splatting Distillation cs.CVPDF

Puyuan Zhang, Jianming Huang, Wenkai Ye, Wei Dong

TL;DR: 本文提出了一种名为G^2ARD-GS的几何引导蒸馏方法，用于将稠密的3D高斯点云表示压缩为紧凑、可重用的模型。该方法通过渐进式整合表面感知代表点，并在构建时锚点约束下恢复外观，实现了在保持几何精度的同时大幅减少基元数量。

Details

Motivation: 稠密彩色LiDAR地图生成的3D高斯溅射模型包含数百万基元，导致存储、传输、渲染和适应成本高昂。现有激进压缩方法会破坏局部表面支撑，影响新视角合成和下游几何应用的稳定性。

Result: 在MatrixCity数据集上，G^2ARD-GS在5倍至30倍压缩预算下取得了最佳的PSNR、SSIM和LPIPS指标，PSNR比PUP方法高出3.2-6.8 dB。在剑桥KingsCollege数据集上，30倍压缩时仍保持了图像到模型的配准精度。

Insight: 创新点在于几何引导的锚点正则化蒸馏框架，通过表面感知代表点整合和固定拓扑下的外观恢复，实现了压缩与几何保真度的平衡。其几何感知视图选择机制在有限监督下优化了视图预算分配，提升了模型的可重用性和适应性。

Abstract: Dense colored LiDAR maps provide accurate city-scale geometry, but lifting them into 3D Gaussian Splatting (3DGS) retains millions of primitives, making the resulting models costly to store, transmit, render, and adapt. Aggressive primitive reduction alleviates this burden, but can remove the local surface support needed for stable novel-view synthesis and downstream geometric use. We introduce G$^2$ARD-GS, a geometry-guided distillation method that converts a dense Gaussian prior instantiated either as a training-free point-cloud lift or a trained GS model into a compact, reusable representation. G$^2$ARD-GS progressively consolidates the prior into surface-aware representatives, then recovers appearance on the resulting fixed topology under construction-time anchor constraints, with no primitives added or removed during recovery. Under limited supervision, geometry-aware view selection allocates the available view budget. On MatrixCity, G$^2$ARD-GS achieves the best PSNR, SSIM, and LPIPS across matched $5\times$–$30\times$ compression budgets, outperforming PUP by $3.2$–$6.8$,dB in PSNR. When reused as frozen geometry, the compact model improves off-trajectory appearance adaptation by $3.7$–$4.9$,dB over PUP 3D-GS and preserves image-to-model registration accuracy on Cambridge KingsCollege at $30\times$ compression. Project page: https://patrick1159.github.io/gardGS-page/.

[55] LAWM-3D: Learning 3D-Aware Latent Actions from Human Videos for Generalizable Robot World Models cs.CVPDF

Jiarui Yang, Jiale Zhange, Jiawei Li, Hang Guo, Wen Huang

TL;DR: 本文提出了LAWM-3D，一种从多视角人类视频中学习3D感知潜在动作表示的方法，用于构建可泛化的机器人世界模型。该方法通过多视角不变的动作标记化、几何对齐约束和RGB-D联合重建目标，解决了现有潜在动作模型在2D像素空间的局限性，并在大规模预训练与机器人微调的两阶段范式下验证了其有效性。

Details

Motivation: 现有世界模型在开放世界具身智能中的应用受限于高昂的动作标注成本和跨平台动作空间的异构性。虽然潜在动作模型（LAMs）能从无标注人类视频中自监督学习动作表示，但它们大多依赖单视角输入并在2D像素空间操作，无法自然获得3D感知能力。

Result: 大量实验表明，所提出的3D感知潜在动作显著提升了世界模型的性能，在生成质量、物理一致性和泛化能力方面均达到了最先进（SOTA）水平。

Insight: 创新点在于三个紧密耦合的设计：多视角不变的动作标记化方案、锚定于预训练3D基础模型的几何对齐约束、以及防止未来帧外观信息捷径学习的非单射RGB-D联合重建目标。这些设计共同迫使模型关注具有几何意义的运动线索，从而学习到真正的3D感知动作表示。

Abstract: World models enable agents to perform forward rollout and planning without real-world interaction. However, their application in open-world embodied intelligence remains limited by the high cost of action annotations and the heterogeneity of action spaces across platforms. Recently, latent action models (LAMs) have alleviated this bottleneck by learning action representations directly from unlabeled human videos in a self-supervised manner. Nevertheless, most existing LAMs rely on single-view inputs and operate primarily in 2D pixel space, raising a fundamental question: can simply incorporating multi-view videos into LAM training endow the learned latent actions with 3D-aware perception? Our study shows that the answer is negative. The primary reasons lie in future-frame appearance leakage as well as inter-camera appearance discrepancies and viewpoint variations. To address these issues, we propose LAWM-3D, which introduces three tightly coupled key designs: (1) a multi-view invariant unified action tokenization scheme for learning 3D-aware latent actions; (2) a geometric alignment constraint that anchors intermediate encoder features to a pretrained 3D foundation model, thereby explicitly providing cross-view geometric correspondences; and (3) a non-injective RGB-D joint reconstruction objective that prevents shortcut learning from future-frame appearance information, forcing the LAM to focus supervision on motion cues with geometric significance. Importantly, these components are not simply stacked but are tightly coupled through a unified motivation. Built upon a two-stage paradigm of large-scale human video pretraining followed by robot fine-tuning, extensive experiments demonstrate that the proposed 3D-aware latent actions significantly improve world model performance, achieving SOTA results in generation quality, physical consistency, and generalization ability.

[56] One Ranking, Any Budget: Matryoshka Evidence-to-Context Frame Selection for Long-Video Understanding cs.CVPDF

Wang Chen, Yu Chen, Xiang Wang, Shuai Li, Jinfa Huang

TL;DR: 本文提出了一种名为Matryoshka Evidence-to-Context（MEC）的免训练框架，用于长视频理解中的帧选择。该框架通过构建一个可重用的稀疏视频索引，并采用贪心算法生成一个位置自适应的优先级序列，使得该序列的前缀能集中查询相关的证据，而更大的前缀则在保留证据的同时增加更广泛的时间上下文。因此，单个排名序列可以被截断以适配任何目标预算，无需重新运行选择器。

Details

Motivation: 为了解决长视频中严重的帧冗余和有限上下文窗口问题，现有帧选择方法通常为每个预定义预算优化独立的帧子集，当预算变化时，先前选择的证据可能被替换而非逐步扩充。本文旨在设计一个实用的选择器，能够服务于多种预算，并允许跨预算重用前缀。

Result: 在四个基准测试和六个帧预算下，MEC相比均匀采样将平均准确率提高了3.77个百分点，与强大的最先进（SOTA）选择器性能相当，并将端到端选择延迟降低了47.37-51.19%。

Insight: 创新点在于将长视频帧选择形式化为一个Matryoshka排序问题，构建一个单一优先级序列，其小前缀集中查询条件证据，而更大的前缀则保留证据并增加更广泛的时序上下文。从客观角度看，其免训练、可重用索引和位置自适应贪心排序的设计，在保证性能的同时显著降低了计算开销和延迟，具有很高的实用性。

Abstract: Frame selection is essential for applying Large Multimodal Models (LMMs) to long videos due to severe frame redundancy and limited context windows. Since the appropriate frame budget varies with the downstream LMM, reasoning demands, and latency constraints, a practical selector should serve multiple budgets. However, existing methods typically optimize an isolated frame subset for each predefined budget: when the budget changes, previously selected evidence may be replaced rather than progressively augmented. Ranking frames by a fixed score would allow prefix reuse across budgets, but it ignores the distinct roles of different ranking positions. In this paper, we formulate long-video frame selection as a Matryoshka ranking problem: constructing a single priority sequence whose small prefixes concentrate query-conditioned evidence, while progressively larger prefixes preserve this evidence and add broader temporal context. Efficiently constructing such a ranking is itself challenging, as densely sampling long videos and evaluating frame-query relevance incurs substantial overhead. We therefore introduce Matryoshka Evidence-to-Context (MEC) Frame Selection, a training-free framework that builds a reusable sparse video index, discovers candidates through sparse probing and local zooming, and greedily constructs a position-adaptive ranking: early positions emphasize evidence; later positions progressively favor temporal coverage while preserving visual diversity. A single ranking can thus be truncated to any target budget without rerunning the selector. Across four benchmarks and six frame budgets, MEC improves average accuracy over uniform sampling by 3.77 percentage points, matches strong state-of-the-art selectors, and reduces end-to-end selection latency by 47.37-51.19%.

[57] Engram-E2VID: Reference-Based Event-to-Video Reconstruction via Generative Activation of Appearance Engrams cs.CV | cs.LG | physics.opticsPDF

Feiyu Ji, Xiang Li, Hao Ma, Tianxiang Huang, Qingxin Lu

TL;DR: Engram-E2VID是一个基于参考帧和事件流重建目标RGB视频帧的框架。它通过将参考帧编码为外观记忆单元，并利用事件流生成目标时刻的运动结构支架，在一个一步扩散模型中引导结构令牌激活相关的外观记忆，从而解决事件数据稀疏和缺乏绝对外观信息带来的重建挑战。

Details

Motivation: 解决基于参考帧的事件到视频重建中，如何将事件流捕获的精细时间结构与参考帧中的外观信息有效关联的核心难题，尤其是在复杂运动和长时间间隔下。

Result: 在三个基准测试上，Engram-E2VID相比最强的同输入基线，PSNR提升最高达3.29 dB，LPIPS降低最高达0.08，并且在重建间隔增大时性能下降更慢。

Insight: 创新点在于提出了结构引导的生成激活框架，将外观与结构在令牌空间解耦关联，避免了依赖像素级对应；并利用扩散先验来补全不确定或新出现的区域，增强了重建的鲁棒性和保真度。

Abstract: Reference-based event-to-video reconstruction aims to recover target RGB frames from a reference frame and the event stream captured over the reference-to-target interval. Although events provide fine-grained temporal cues, they encode sparse and asynchronous log-intensity changes rather than absolute appearance, making faithful reconstruction intrinsically challenging. The central challenge lies in associating event-derived target-time structures with relevant appearance information from the reference frame, especially under complex motion and long temporal intervals. In this work, we propose Engram-E2VID, a structure-guided framework that reconstructs target frames through the generative activation of appearance engrams. Specifically, the reference frame is encoded into token-space appearance engrams, while the event stream and reference context are transformed into a target-time motion-structure scaffold that captures motion boundaries and event-induced structural changes. Within a one-step diffusion backbone, scaffold-derived structural tokens progressively interact with and activate relevant appearance engrams across layers. This token-space association allows target structures to access reference appearance without relying on direct pixel-wise correspondence, while the diffusion prior complements uncertain or newly revealed regions. Across three benchmarks, Engram-E2VID improves PSNR by up to 3.29 dB and reduces LPIPS by up to 0.08 over the strongest same-input baseline, while degrading more slowly as the reconstruction interval increases.

[58] Beyond Relevance: Bayesian Evidence Acquisition for Agentic Whole-Slide Image Reasoning cs.CVPDF

Bryan Wong, Xun Xu, Huazhu Fu, Nancy F. Chen, Mun Yong Yi

TL;DR: 本文提出BEACON，一种即插即用的代理框架，将全切片图像（WSI）推理重新表述为贝叶斯证据获取问题。该框架无需额外训练，通过最大化期望信息增益来顺序获取图像块以减少诊断不确定性，并在五个WSI-VQA基准测试中实现了零样本下的最佳整体性能和更高的证据获取效率。

Details

Motivation: 现有基于语义相关性的迭代块检索方法在计算病理学中存在问题，因为语义相关并不一定意味着具有诊断信息性，导致可能获取冗余观察且无法解决诊断不确定性。

Result: 在五个WSI-VQA基准上进行的大量零样本实验表明，BEACON在无需训练的代理框架中实现了最强的整体性能，并显著提高了证据获取效率。

Insight: 创新点在于将WSI推理形式化为贝叶斯证据获取问题，通过维护对竞争诊断假设的概率信念并最大化期望信息增益来主动减少不确定性，为不确定性感知的代理WSI推理提供了一个原则性范式。

Abstract: Whole-slide image (WSI) reasoning requires an agent to sequentially acquire visual evidence before answering a diagnostic question. Existing training-free agentic frameworks formulate this process as iterative patch retrieval based on semantic relevance to the question. However, semantic relevance does not necessarily imply diagnostic informativeness in computational pathology, where competing diagnoses often exhibit similar and overlapping morphological patterns, making many patches semantically relevant yet diagnostically non-discriminative. Consequently, relevance-based retrieval may acquire redundant observations and leave diagnostic uncertainty unresolved. We propose BEACON, a plug-and-play agentic framework that reformulates WSI reasoning as a Bayesian evidence acquisition problem. BEACON maintains a probabilistic belief over competing diagnostic hypotheses and sequentially acquires patches by maximizing expected information gain (EIG) to reduce diagnostic uncertainty. An evidence controller then determines whether to answer, acquire additional evidence, or perform higher-resolution inspection. Built entirely from off-the-shelf foundation models, BEACON requires no additional training or fine-tuning. Extensive zero-shot experiments across five WSI-VQA benchmarks demonstrate that BEACON achieves the strongest overall performance among training-free agentic frameworks while substantially improving evidence acquisition efficiency, establishing Bayesian evidence acquisition as a principled paradigm for uncertainty-aware agentic WSI reasoning. The code is available at https://github.com/bryanwong17/BEACON

[59] UniVVT: A Unified End-to-End Framework for High-Fidelity Video Virtual Try-on cs.CV | cs.AIPDF

Yushe Cao, Shikun Feng, Fei Shen, Haikuo Peng, Jianqiang Xia

TL;DR: 本文提出了UniVVT，一个用于高保真视频虚拟试穿的统一端到端框架。它将VVT任务重新定义为语义条件视频生成，摒弃了传统方法中依赖的掩码、姿态和服装变形等显式几何先验模块。其核心是基于多模态大语言模型构建的场景任务感知器，将源视频、目标服装和任务指令联合编码为紧凑的任务感知潜在令牌，并通过轻量级语义桥与扩散视频生成器的条件空间对齐。

Details

Motivation: 当前主流的视频虚拟试穿方法采用多阶段设计，依赖独立的人体解析、姿态估计和服装变形模块。这种设计不仅部署复杂，更关键的是，显式几何先验中的误差会不可逆地传播到生成的视频中。本文旨在消除这些模块，构建一个更鲁棒、统一的端到端框架。

Result: 广泛的实验表明，UniVVT在多个基准测试中达到了最先进的性能。

Insight: 主要创新点在于将VVT任务重新定义为语义条件视频生成，并引入了基于多模态大语言模型的场景任务感知器来隐式捕获“转移什么”、“在哪里转移”以及“如何转移”的语义信息。这为端到端虚拟试穿提供了一种简单有效的替代方案，避免了脆弱的几何预处理。从客观角度看，其提出的三阶段渐进式训练策略（语义对齐、联合任务适应和灵活分辨率细化）对于有效耦合异构组件也颇具借鉴意义。

Abstract: Video Virtual Try-On (VVT) synthesizes a video of a person wearing a target garment while preserving identity, motion, and scene dynamics. Dominant approaches cast VVT as mask-conditioned video inpainting and rely on separate modules for human parsing, pose estimation, and garment warping. This multi-stage design complicates deployment and, more critically, allows errors in explicit geometric priors to propagate irreversibly into the generated video. We present UniVVT, a unified end-to-end framework that reframes VVT as semantically conditioned video generation, eliminating mask, pose, and warping modules at inference. At its core, a scene-task perceiver built on a Multimodal Large Language Model jointly encodes the source video, target garment, and task instruction into compact, task-aware latent tokens, implicitly capturing what to transfer and where and how to transfer it. A lightweight semantic bridge then aligns these tokens with the conditioning space of a diffusion-based video generator, enabling coherent garment transfer. To robustly couple the heterogeneous components, we devise a three-stage progressive training strategy comprising semantic alignment, joint task adaptation, and flexible-resolution refinement. Extensive experiments demonstrate that UniVVT achieves state-of-the-art performance across multiple benchmarks, validating implicit semantic guidance as a simple and effective alternative to fragile geometric preprocessing for end-to-end virtual try-on.

[60] GST-Bench: Can VLMs Develop Global Spatial Awareness from Video? cs.CVPDF

Qifeng Zhang, Kaixiang Huang, Heng Dong, Huang Fang, Junting Chen

TL;DR: 本文提出了GST-Bench，一个用于评估视频语言模型全局空间智能的VQA基准。该基准包含基于大量合成视频生成的人类验证问题，要求模型从输入视频中未见的新视角进行准确的空间推理，并将第一人称观察映射到全局俯视图像上。对22个最先进模型的评估揭示了模型与人类表现之间的巨大差距，并进一步通过GST-Bench-Local分析发现模型难以将长时程观察整合为全局一致的场景表示。

Details

Motivation: 现有基准主要关注从单一或少数视角的局部空间感知，而忽视了在连续、长时程视觉流上的全局空间意识。为了弥补这一局限性，本文旨在评估和推动视频语言模型发展全局空间智能。

Result: 在GST-Bench上，最强的零样本模型仅获得42.68分，远低于人类得分79.08。在构建的GST-Bench-Local上，模型表现出较强的局部空间理解能力，但仍无法整合长时程观察形成全局一致的场景表示。

Insight: 论文的核心创新点是提出了首个专注于评估视频语言模型全局空间推理能力的基准GST-Bench，其任务设计（如从新视角推理、映射到全局地图）直接针对现有模型的短板。同时，通过构建局部对照基准，深入剖析了模型失败的原因在于整合长时程信息形成全局表征的困难，而非局部感知能力不足，这为未来研究指明了方向。提供的GST-Train数据集也为解决这一挑战提供了资源。

Abstract: Spatial intelligence is fundamental to embodied agents, yet existing benchmarks focus on local spatial perception from single or few viewpoints, overlooking global spatial awareness over continuous, long-horizon visual streams. To address this limitation, we introduce the Global-Spatial-Temporal Benchmark (GST-Bench), a VQA benchmark for global spatial intelligence in video understanding, comprising human-verified questions derived from 6,790 minutes of synthetically generated video. It requires models to perform accurate spatial inference from novel viewpoints unseen in the input video and to map egocentric observations onto global top-down images. A comprehensive evaluation of 22 state-of-the-art VLMs exposes a striking gap between models and humans: the strongest zero-shot model attains only 42.68, far below the human score of 79.08. To probe the cause of this gap, we construct GST-Bench-Local and find that models, despite strong local spatial understanding under the same task formulation, still fail to consolidate long-horizon observations into a globally consistent scene representation. We further provide GST-Train, a dataset for global spatial reasoning, as a complementary resource to facilitate future research on this challenge.

[61] Flow-Map Distillation on Relation Manifolds for Image Restoration cs.CVPDF

Zihao He, Songhua Liu

TL;DR: 本文提出了一种名为FoRM（Flow-Map Distillation on Relation Manifolds）的知识蒸馏方法，用于图像恢复任务。该方法将基于关系的知识传递重新定义为关系流形上的连续流映射问题，通过学习一个流映射算子来预测任意目标时刻的关系状态，从而提供轨迹级的监督。

Details

Motivation: 现有图像恢复知识蒸馏方法通常将师生网络中间特征或关系矩阵作为静态目标对齐，忽略了知识传递过程的动态结构。本文旨在解决这一局限性。

Result: 在超分辨率、去雨、去噪、去模糊和低光增强五个图像恢复任务上的大量实验表明，该方法在多种骨干网络上均优于最先进的蒸馏基线，训练方差降低约50%，并实现了更优的恢复质量。

Insight: 创新点在于将知识蒸馏建模为关系流形上的连续流映射问题，并引入了安全半群一致性约束和端点锚定损失来确保流映射的全局自一致性和目标对齐，避免了幻象状态误差累积。

Abstract: Knowledge distillation for image restoration typically aligns intermediate features or relation matrices between teacher and student networks as static targets, ignoring the dynamic structure of the knowledge transfer process. In this paper, we propose Flow-Map Distillation on Relation Manifolds (FoRM), which reformulates relation-based knowledge transfer as a continuous flow mapping problem on the relation manifold. Rather than regressing a constant velocity field between student and teacher relation states, FoRM learns a flow map operator $\mathcal{F}_θ(\mathbf{z}, t, s)$ that directly predicts the relation state at any target time $s$ given the current state at time $t$, enabling richer trajectory-level supervision. To ensure global self-consistency of the learned flow map, we introduce a safe semigroup consistency constraint that enforces compositional agreement using ground-truth bridge states, eliminating phantom-state error accumulation. An endpoint anchoring loss further prevents the operator from drifting away from the teacher target. Extensive experiments on five image restoration tasks, including super-resolution, deraining, denoising, deblurring, and low-light enhancement, demonstrate consistent gains over state-of-the-art distillation baselines across multiple backbone architectures, reducing training variance by approximately 50% compared to naive flow matching distillation while achieving superior restoration quality.

[62] HyTBE: Hyperbolic Target-Background Expert Model for Cross-Domain Infrared Small Target Detection cs.CV | cs.AIPDF

Aohua Li, Jin Kuang, Yubing Lu, Pingping Liu

TL;DR: 本文提出HyTBE模型，一种用于跨域红外小目标检测的双曲目标-背景专家模型。该模型通过目标-背景关系干预扩展源域关系模式，并利用双曲空间建模目标-背景关系，通过MoE适配器自适应校准视觉特征，以解决因目标-背景关系偏移导致的跨域性能下降问题。

Details

Motivation: 现有红外小目标检测方法在域一致评估下取得进展，但在泛化到未见红外域时性能显著下降。作者将此归因为目标-背景关系偏移：训练时仅从有限源域关系模式学习决策规则，无法适应未见域的新关系模式。

Result: 在NUAA-SIRST、NUDT-SIRST和IRSTD-1K数据集上进行的留一域出实验表明，HyTBE相比竞争基线实现了更强的跨域泛化能力。

Insight: 创新点在于将跨域失败形式化为目标-背景关系偏移问题，并引入双曲空间（Poincaré球）建模目标与背景的相对距离关系。通过关系干预扩展训练模式，并利用基于双曲关系的MoE适配器进行特征校准，为跨域小目标检测提供了新的关系感知框架。

Abstract: Infrared small target detection (IRSTD) has achieved substantial progress under domain-consistent evaluation, yet detector performance often degrades markedly when generalizing to unseen infrared domains. Existing methods primarily improve detection by enhancing target responses and suppressing background interference. However, when trained on only a limited set of source domains, their learned decision rules are inevitably established from a restricted range of source-domain target-background relation patterns. We formulate this cross-domain failure as target-background relation shift: unseen domains may exhibit relation patterns that are not observed during training, thereby weakening the discriminative capability learned from the source domains. To address this problem, we propose HyTBE, a Hyperbolic Target-Background Expert model that expands source-domain relation patterns and adaptively adjusts visual representations using explicit relation cues. The Target-Background Relation Intervention selectively perturbs either targets or backgrounds, broadening the observable relation patterns during training while maintaining valid supervision. Subsequently, the Hyperbolic Relation Modeling maps multi-scale visual cues into a Poincaré ball and characterizes the target-background relation of each feature token according to its relative distances to the target and background anchors. The Hyperbolic-guided MoE Adapter further uses these hyperbolic relation representations to calibrate multi-scale visual features and aggregate expert-specific feature corrections for different relation patterns. Leave-one-domain-out experiments on NUAA-SIRST, NUDT-SIRST, and IRSTD-1K demonstrate that HyTBE achieves stronger cross-domain generalization than competitive baselines.

[63] Evidence-Driven Dynamic Visual Selector for Efficient Long Video Understanding cs.CVPDF

Bo Zhang, Wenxin Wang, Feng Chen, Zhihao Zhang, Zixuan Wang

TL;DR: 本文提出了一种名为EviSelect的细粒度动态视觉选择框架，用于高效的长视频理解。该方法基于目标多模态大语言模型（MLLM）的内部注意力证据，通过稀疏预填充来高效探测视觉证据作为结构化先验，以指导分布感知的动态采样。具体而言，它利用高度压缩的视觉输入和稀疏注意力来高效近似目标MLLM的注意力图，并基于此先验的三个互补注意力组件设计了一个轻量级选择器，该选择器不仅能精确定位查询相关的时间戳，还能自适应地调整局部采样率和空间分辨率。

Details

Motivation: 现有的基于MLLM的长视频理解方法通常依赖外部代理评分器和僵化的启发式规则来选择查询相关帧，这不可避免地与目标MLLM的内在证据不匹配，且无法适应非均匀的时空信息密度。

Result: 在三个长视频理解基准测试上，EviSelect相比现有方法实现了更优的性能，同时将选中的视觉令牌减少了约50%，并实现了3.9倍的端到端加速。

Insight: 创新点在于提出了一种基于目标MLLM内部注意力证据的动态视觉选择框架，通过将选择器建模为随机策略，并在联合准确率-效率奖励下通过GRPO进行优化，实现了根据视频信息密度动态分配计算资源。这避免了对外部代理的依赖，并更好地适应了视频内容的非均匀性。

Abstract: Recent advancements in MLLM-based long-form video understanding have mitigated inference-time computational cost and limited context lengths by selecting query-relevant frames. However, existing approaches predominantly rely on external proxy scorers and rigid heuristic rules, inevitably suffering from misalignment with the target MLLM’s intrinsic evidence and failing to accommodate the non-uniform spatiotemporal information density. In this paper, we propose a fine-grained dynamic visual selection framework named EviSelect, grounded in the target MLLM internal attention evidence. Our method efficiently probes visual evidence via sparse prefilling as a structured prior to guide distribution-aware dynamic sampling. Specifically, we efficiently approximate attention maps of the target MLLM using highly compressed visual inputs and sparse attention, well-aligned to the full counterpart. Conditioned on three complementary attention components derived from this prior, we design a lightweight selector that not only precisely locates query-relevant timestamps but also adaptively adjusts the local sampling rate and spatial resolution. To enable evidence-conditioned spatiotemporal sampling, we formulate the selector as a stochastic policy and optimize it via GRPO under a joint accuracy–efficiency reward. By rewarding correct predictions under lower visual cost through group-relative comparisons, our method encourages the policy to allocate computation dynamically according to the information density of each video. Across three long video understanding benchmarks, EviSelect achieves superior performance compared to existing methods while reducing selected visual tokens by about 50% and achieving a 3.9x end-to-end speedup.

[64] Vorch-Director: Interactive World Story Model via Noise-Aware Error Rectification cs.CVPDF

Lisai Zhang, Yidi Wu, Qi Liu, Xin Ma, Yang Ding

TL;DR: 本文提出Vorch-Director，一种用于长时视听生成的噪声感知残差校正策略。它通过将训练中注入的合成误差与去噪过程的噪声水平对齐，缓解了自回归生成中因历史误差累积导致的身份漂移、过度平滑和音画不同步问题。该方法基于LTX-2扩散变换器，通过任务嵌入和混合任务训练，支持多镜头、多主体、参考引导的长视频生成。

Details

Motivation: 自回归延续方法在长时视听生成中，模型在训练时使用干净的真实历史数据，而在推理时依赖自身生成的、带有累积误差的历史数据，这导致了身份漂移、过度平滑和音画不同步等问题。现有方法通过重用预测残差作为合成噪声来减少这种不匹配，但其效果严重依赖于产生残差时的流匹配噪声水平。

Result: 在ST-Bench和新提出的长时视听基准测试上进行了广泛实验，结果表明Vorch-Director在稳定性和音画保真度方面优于强基线模型。新基准包含质量漂移和长程一致性的评估指标。

Insight: 核心创新是提出了一种噪声水平感知的残差校正策略，将每个残差与其来源的噪声水平关联，并在训练时从匹配的噪声机制中注入残差，从而将注入的误差与去噪过程对齐。此外，引入了任务嵌入来区分历史视频、参考图像和目标视频，实现了长时生成任务的统一条件化，并结合了干净条件汇和混合任务训练以支持复杂的生成场景。

Abstract: Autoregressive continuation provides a natural path toward minute-scale audio-visual generation by repeatedly extending a short-window generator conditioned on previously generated video and audio. However, models are trained on clean ground-truth histories, while inference relies on their own generated histories, where accumulated errors cause identity drift, over-smoothing, and audio-visual desynchronization. Recent methods reduce this mismatch by reusing prediction residuals as synthetic corruption, but we observe that the effectiveness of residual correction critically depends on the flow-matching noise level at which residuals are produced. We propose Vorch-Director, a noise-level-aware residual correction strategy that associates each residual with its originating noise level and injects residuals from matched noise regimes during training. By aligning injected errors with the denoising process, Vorch-Director produces more realistic autoregressive histories while retaining efficient teacher-forcing training. Built on the audio-visual LTX-2 diffusion transformer, Vorch-Director further introduces task embeddings to distinguish historical video, reference images, and target video, enabling unified conditioning for long-horizon generation. Together with a clean conditioning sink and mixed-task training, Vorch-Director supports multi-shot, multi-subject, reference-guided audio-visual long-video generation. We evaluate Vorch-Director on ST-Bench and introduce a new long-horizon audio-visual benchmark with metrics for quality drift and long-range consistency. Extensive experiments demonstrate improved stability and audio-visual fidelity over strong baselines.

[65] VSMP-IMU: Video-Grounded Semantic Motion Programs for Sensor-Aware Synthetic IMU Generation cs.CV | cs.LGPDF

Lala Shakti Swarup Ray, Vitor Fortes Rey, Mengxi Liu, Paul Lukowicz, Bo Zhou

TL;DR: 论文提出了VSMP-IMU框架，用于生成可穿戴设备相关的合成IMU数据。该框架基于视频驱动的语义运动程序，将活动定义语义与标签保持的变体分离，通过提取、增强SMP、合成运动并转换为虚拟IMU信号，最终将信号适配到目标可穿戴领域，以解决可穿戴人类活动识别中标记传感器数据稀缺的问题。

Details

Motivation: 可穿戴人类活动识别常受限于标记传感器数据的稀缺性，尤其是在低资源、类别不平衡和主体泛化场景中。现有的合成IMU生成方法存在权衡：视频驱动方法视觉基础好但对姿态估计误差敏感，文本驱动方法可控但往往与实际活动执行方式关联较弱。

Result: 在五个公共IMU-HAR数据集上，采用留一人出评估，VSMP-IMU的平均Macro-F1达到78.33%，比仅使用真实数据训练提高了9.77%，比先前最强的合成基线提高了4.04%。在低资源设置下，平均提升超过6%；在不平衡数据集的长尾评估中，尾部类别的Macro-F1比仅使用真实数据训练提高了19.86%，比SOTA提高了4.76%。

Insight: 创新点在于引入了结构化的、视频基础的语义运动程序，实现了可控且与可穿戴领域相关的合成传感器数据生成。客观分析认为，该方法通过分离活动语义与变体，增强了生成数据的实用性和基础性，有效平衡了视觉基础与可控性之间的矛盾。

Abstract: Wearable human activity recognition (HAR) is often limited by the scarcity of labeled sensor data, especially in low-resource, class-imbalanced, and subject-generalization settings. Synthetic IMU generation can reduce this dependency and enhance HAR machine learning model’s performance, but existing approaches face a trade-off without addressing all factors: video-driven methods are visually grounded but sensitive to pose-estimation errors, while text-driven methods are controllable but often weakly grounded in how activities are actually performed. We present VSMP-IMU, a video-grounded framework for controllable synthetic IMU generation based on a structured Semantic Motion Program (SMP), which separates activity-defining semantics from label-preserving variation. Given an input video, VSMP-IMU extracts and augments an SMP, uses it to synthesize motion, converts the motion into virtual IMU signals, and grounds the resulting signals to the target wearable domain. We evaluate VSMP-IMU against state-of-the-art synthetic data generation methods on five public IMU-HAR datasets under leave-one-person-out evaluation. VSMP-IMU achieves an average Macro-F1 of 78.33%, improving over real-only training by 9.77% and over the strongest prior synthetic baseline by 4.04%. In low-resource settings with reduced training data-samples, it improves over real-only training by 18.54% and over the strongest prior synthetic baselines by more than 6% on average. Under long-tail evaluation in imbalanced datasets, it improves tail-class Macro-F1 by 19.86% over Real-only training and by 4.76% over SOTA. These results show that structured video-grounded semantics provide a practical foundation for controllable, wearable-relevant synthetic sensor data generation.

[66] KVAE: Family of Tokenizers for Multimodal Generative Models cs.CV | cs.LG | cs.SDPDF

Andrey Shutkin, Denis Parkhomenko, Ivan Kirillov, Kirill Chernyshev, Kirill Malakhov

TL;DR: 本文介绍了KVAE系列分词器，包括KVAE-Audio（音频）、KVAE-3D（视频）和KVAE-2D（图像），它们专为多模态生成模型设计，用于将输入信号压缩为潜在表示。这些分词器在重建质量和生成性能上达到或超越了当前前沿的开源分词器，如Wan-2.2、HunyuanVideo-1.5等，并公开了代码和训练细节。

Details

Motivation: 潜在扩散模型（LDM）依赖分词器将输入信号映射到压缩表示，分词器影响学习速度、生成样本质量及后续应用，因此需要开发高效的多模态分词器以支持文本条件生成。

Result: 在重建指标（如PSNR、LPIPS、PESQ）和生成指标（如Frechet距离、CLIP分数、CLAP分数）上，KVAE分词器在客观和主观评估中匹配或超越了前沿开源分词器，适用于音频、图像和视频模态。

Insight: 创新点在于设计了针对不同模态（音频、视频、图像）的专用分词器架构，如连续全频带音频压缩和因果视频压缩，并通过公开训练细节和模型选择方法促进了社区发展，提升了多模态生成模型的效率和性能。

Abstract: Latent diffusion modeling (LDM), a prominent paradigm, utilizes tokenizers to map input signal to compressed representation. This dependency positions tokenizer as an integral part of generation process itself, since it affects learning speed, quality of synthesized samples and lay foundation for later applications. This report presents series of KVAE tokenizers for audio, image and video, all designed for subsequent text-conditioned generation: KVAE-Audio, a continuous full-band 48 kHz tokenizer with a 50 Hz latent of 64 channels; KVAE-3D – two causal video tokenizers for 4x16x16 and 4x8x8 compression; KVAE-2D, an image model, compressing input by factor of 8 with 32 channels. We demonstrate that reconstruction (PSNR, LPIPS, PESQ, etc.) and generation results on objective (Frechet Distance, CLIP score, CLAP score, etc.) and subjective (side-by-side evaluation) metrics matches or surpasses frontier opensource tokenizers, such as VAEs from Wan-2.2, HunyuanVideo-1.5, FLUX.2, MovieGen, StableAudio and MMAudio. Considering difficulty of development, we share with community training details, model selection method and ablation on design choices. The code is publicly available at https://github.com/kandinskylab/kvae and https://github.com/kandinskylab/kvae-audio.

[67] Vorch-Omni: Multi-Task Orchestration of Sight and Sound cs.CVPDF

Vorch Team, Xiaoyu Chen, Yang Ding, Cong Han, Menglin Han

TL;DR: 本文提出了Vorch-Omni，一个基于任意条件到任意输出公式的统一多任务视听合成框架。该框架将视频和音频信号灵活地视为条件输入或生成目标，通过令牌级条件掩码和任务标识符来区分目标、源内容和参考信号，并采用互补的视觉条件路径来捕获语义和结构信息。基于单一流匹配扩散Transformer构建，无需任务特定的架构更改，即可支持超过10项任务，为通用视听生成和操作提供了可扩展的基础。

Details

Motivation: 现有生成视频建模方法通常依赖碎片化的任务特定模型，缺乏通用性。一个通用模型需要区分异构的目标、源和参考信号，以确定生成、保留或指导的内容，同时减少任务间的干扰，而联合视听生成进一步增加了这一挑战。

Result: 论文构建了一个分布式数据管道来策划多样化的时间对齐视听片段，并生成结构化描述和元数据以平衡异构任务分布。所提出的Vorch-Omni框架支持超过10项任务，包括文本到视频、文本到视听、图像和参考条件生成、时间扩展、音频驱动生成、视频转换和视听编辑等。

Insight: 创新点在于提出了一个统一的任意条件到任意输出框架，通过令牌级条件掩码、任务标识符和位置类型来灵活处理多模态信号，并采用互补的视觉条件路径（视觉语言模型和视频VAE）来融合语义和结构信息。其基于单一扩散Transformer的架构设计避免了任务特定的修改，实现了多任务统一建模的可扩展性。

Abstract: Recent advances in generative video modeling have enabled diverse generation, reference-based synthesis, extension, and editing, but existing approaches often rely on fragmented task-specific models. A general model must distinguish heterogeneous target, source, and reference signals to determine what to generate, preserve, or use as guidance, while reducing interference among tasks. Joint audio-visual generation further increases this challenge by introducing diverse conditioning and output configurations across modalities. We present Vorch-Omni, a unified multi-task framework for audio-visual synthesis based on an arbitrary-condition-to-arbitrary-output formulation. It flexibly treats video and audio signals as either conditioning inputs or generation targets. Token-level conditioning masks and task identifiers distinguish targets, source content, and references, while position types separate temporal context from independent conditions. To capture semantic and structural information, Vorch-Omni employs complementary visual conditioning pathways: a vision-language model interprets sampled frames with text instructions, and a video VAE encodes conditions into latent tokens for direct guidance. We further build a distributed data pipeline to curate diverse temporally aligned audio-visual clips, generate structured captions and metadata, and balance heterogeneous task distributions. Built on a single flow-matching diffusion transformer without task-specific architectural changes, Vorch-Omni supports over 10 tasks, including text-to-video, text-to-audio-video, image- and reference-conditioned generation, temporal extension, audio-driven generation, video transformation, and audio-visual editing. This unified framework provides a scalable foundation for general-purpose audio-visual generation and manipulation.

[68] Overcoming Attention Drift: Homogeneity-Heterogeneity Guided Feature Aggregation for Low-Light Remote Sensing Image Enhancement cs.CVPDF

Yaozi Zhong, Xingxing Yang, Shaohui Mei, Mingyang Ma

TL;DR: 论文提出了一种名为HALO的双先验驱动增强框架，用于解决极端低光条件下遥感图像增强中的注意力漂移问题。该方法通过引入光照不变语义先验和伪3D拓扑先验，分别提供区域同质性正偏置和边界异质性负惩罚，以指导特征聚合，并设计了同质-异质协同注意力模块（H2CAM）来融合这些先验。

Details

Motivation: 现有方法在严重噪声和光照退化下存在注意力漂移问题，导致跨物理边界错误聚合特征，引起严重的结构模糊和颜色失真，因此需要一种能够可靠恢复高保真遥感图像的方法以支持地球观测和下游机器视觉任务。

Result: 在8个具有挑战性的合成和真实世界遥感基准测试中，HALO实现了最先进的性能，显著提高了物理边界清晰度和颜色保真度，同时最大程度地保留了下游地球观测任务所需的判别性特征。

Insight: 创新点在于将增强问题形式化为由基础模型先验驱动的引导特征聚合问题，并协同利用同质性和异质性两种互补先验来纠正注意力漂移；客观来看，其提出的H2CAM模块为解决跨模态先验融合中的特征冲突提供了一种有效机制。

Abstract: Restoring high-fidelity remote sensing imagery from extreme low-light degradation is indispensable for reliable Earth observation and downstream machine vision. However, under severe noise and illumination corruption, existing methods suffer from attention drift, erroneously aggregating features across distinct physical boundaries and causing severe structural blurring and color distortion. To address this, we propose HALO, a dual-prior-driven enhancement framework that formulates enhancement as a guided feature aggregation problem driven by foundation model priors. Specifically, an illumination-invariant semantic prior provides regional homogeneity as a positive bias for content-consistent aggregation, while a pseudo-3D topological prior provides boundary heterogeneity as a negative penalty to strictly prevent cross-boundary confusion. To cooperatively incorporate these two priors, we propose a Homogeneity-Heterogeneity Cooperative Attention Module (H2CAM) to resolve feature conflicts during cross-modal prior fusion. Extensive experiments demonstrate that HALO achieves state-of-the-art performance across 8 challenging synthetic and real-world remote sensing benchmarks, significantly improving physical boundary sharpness and color fidelity while maximizing the preservation of discriminative features for downstream Earth observation tasks.

Théo Danielou, Antoine Saporta, Léo Alberge, Corentin Dancette

TL;DR: 本文提出Curia-MAE，一种用于3D医学图像分割的多模态、多解剖学MAE预训练模型。该方法通过增强的卷积MAE预训练，包括鲁棒重建目标、特征正则器和局部-全局相似性目标，在30万张CT和MRI图像上进行预训练。在八个解剖和病灶分割基准测试中，该模型提升了冻结编码器的性能，并在病灶任务上表现优异。

Details

Motivation: 现有放射学基础模型在密集预测任务（如3D分割）上评估不足，且冻结编码器的预训练模型性能仍不及从头训练的SOTA方法nnU-Net，因此需要改进预训练方法以缩小这一差距。

Result: 在八个解剖和病灶分割基准测试上，Curia-MAE相比强MAE基线提升了冻结编码器的性能；在全微调下保持竞争力，在标注数据稀缺的病灶任务上表现更优。

Insight: 创新点包括引入鲁棒重建目标、特征正则器和局部-全局相似性目标来增强卷积MAE预训练；其核心在于通过多模态、多解剖学的大规模预训练，实现单个冻结编码器在多样分割任务中的可重用性，降低临床部署成本。

Abstract: Radiology foundation models learn transferable representations that can be adapted to new tasks by training only small layers on top of a frozen encoder. Dense prediction tasks such as 3D segmentation are, however, underrepresented in their evaluation, and, with the encoder kept frozen, pre-trained models still fall short of nnU-Net, the state-of-the-art reference trained from scratch. To close this gap we extend convolutional MAE pre-training with a robust reconstruction objective, a feature regularizer, and a local-global similarity objective. Using this method, we propose Curia-MAE, a multi-modal, multi-anatomy MAE model pre-trained on 300,000 CT and MRI images covering a large number of anatomical sites. On eight anatomy- and lesion-focused segmentation benchmarks, Curia-MAE improves frozen-encoder performance over a strong MAE baseline, while remaining competitive under full finetuning and superior on lesion tasks, where labeled data is scarce. These results indicate that a single frozen encoder can be reused across diverse segmentation tasks, reducing the cost of adapting and deploying such models in clinical workflows. We will make our pre-trained model weights publicly available.

[70] D-CLOT: Double Closed Loop Optimal Transport for Unsupervised Action Segmentation cs.CV | cs.AIPDF

Elena Bueno-Benito, Mariella Dimiccoli

TL;DR: 本文提出D-CLOT，一种用于无监督动作分割的双闭环最优传输框架。它在CLOT基础上引入图约束模块来正则化帧和片段表示，并周期性地通过k-means或加权OT重心重新估计动作原型，以解决表示与原型不一致的问题。

Details

Motivation: 现有基于最优传输的无监督动作分割方法中，定义OT成本的动作原型仅通过伪标签损失的梯度更新，而未从细化的帧几何中重新估计，导致表示与原型不一致，尤其在模糊过渡和短/低频动作上成为瓶颈。

Result: 在五个基准测试中，D-CLOT的两个变体均优于CLOT，在YTI上单视频F1提升高达+12.7，mIoU提升+10.2；在FS-Eval上活动级F1提升高达+8.9。此外，在更细粒度的Assembly101基准上建立了首个无监督动作分割基线。

Insight: 创新点在于通过图约束模块保持编码器输出的局部邻域几何来正则化表示，并引入基于k-means或OT重心的原型重估计机制，使原型与当前传输几何保持一致，从而解决了表示-原型不一致问题。

Abstract: Optimal transport (OT) has emerged as an effective framework for unsupervised action segmentation. Yet, in existing OT-based methods, the latent action prototypes that define the OT costs are not re-estimated from the refined frame geometry. Instead, they evolve solely through gradients from the pseudo-label loss. We identify this \emph{representation–prototype inconsistency} as a central bottleneck, particularly around ambiguous transitions and for short or infrequent actions. To address this issue, we build on the recently introduced CLOT, which refines frame embeddings based on estimated segment embeddings, and further re-estimates the action prototypes from the refined frame embeddings. Specifically, we introduce a graph-constrained module that regularizes the OT-refined frame and segment representations by preserving the local neighborhood geometry of the encoder output. An action-embedding refinement step then periodically re-anchors the prototypes to this stabilized representation geometry. We study two instantiations that share the same backbone, graph module, and objective: D-CLOT updates the prototypes using $k$-means, whereas D-CLOT$_{B}$ updates them as OT barycenters weighted by the refined transport plan, yielding an assignment-aware prototype update consistent with the current transport geometry. Across five established benchmarks, both variants improve segment-level quality over CLOT, with per-video gains of up to $+12.7$ F1 and $+10.2$ mIoU (YTI) and activity-level gains of up to $+8.9$ F1 (FS-Eval). We further establish the first unsupervised action-segmentation baseline on Assembly101, a procedural and substantially more fine-grained benchmark than those commonly used in prior work. Extensive ablations and sensitivity analyses demonstrate that the two refinement mechanisms are complementary and robust.

[71] Mapping Armenian Paris: Extracting and Geocoding Commercial Advertisements from the 20th-Century Diaspora Press cs.CVPDF

Chahan Vidal-Gorène, Seda Kirakosyan, Edita Matevosyan

TL;DR: 本文提出了一种基于IIIF的端到端流程，将法国亚美尼亚语报刊数字化内容转化为20世纪巴黎亚美尼亚商业社区的交互式地图。该流程通过视觉语言模型定位、读取并解析商业广告为结构化记录，随后进行地理编码并映射。

Details

Motivation: 解决西亚美尼亚语作为低资源历史语言缺乏现成布局分析和OCR模型支持的问题，旨在从20世纪流散报刊中自动提取商业广告信息并进行地理可视化。

Result: 构建了包含500页西亚美尼亚语报刊的语料库，带有3,270个广告级标注；所提流程在传统CRNN OCR失效的弯曲扫描图像上仍保持可靠性，实现了手工标注无法达到的规模。

Insight: 创新点在于利用视觉语言模型作为数据引导策略处理低资源历史语料，开发了集成检测与语义字段标注的Label Studio模板，其可复现工作流可迁移至其他低资源历史语料库。

Abstract: This paper presents an end-to-end, IIIF-based pipeline that turns the digitised Armenian press of France into an interactive map of the 20th-century Parisian Armenian commercial community. On each page, commercial advertisements are located, read, and parsed into structured records, which are then geocoded and placed on the map. Western Armenian is under-resourced and unsupported by off-the-shelf layout and OCR models, so the pipeline uses vision-language models (VLMs) as a data-bootstrapping strategy: they produce usable structured records at a scale hand annotation could not reach, and stay reliable on the strongly curved scans where conventional line-level CRNN OCR breaks down. The contribution includes a 500-page Western Armenian press corpus with 3,270 advertisement-level annotations, a Label Studio template that captures detection and semantic fields in a single annotation pass, and a reproducible workflow transposable to other under-resourced historical corpora. More broadly, the work shows that VLM-driven data bootstrapping is an effective lever for under-resourced historical languages such as (Western) Armenian.

[72] Robust-WAM: Bridging Generative Pretraining and Semantic Foresight in World-Action Models cs.CV | cs.ROPDF

Haodong Yan, Junfeng Li, Junjie He, Zhide Zhong, MingMing Yu

TL;DR: 本文提出Robust-WAM，一种用于视频生成世界动作模型（WAMs）的后训练方法，旨在解决现有模型在视觉分布外条件下鲁棒性不足的问题。该方法在保留基于VAE的生成路径的同时，通过轻量级的语义前瞻对齐目标，将动作流与外观不变的动态语义进行对齐，从而在光照变化等视觉偏移下保持可靠性。

Details

Motivation: 主流世界动作模型（WAMs）通常基于变分自编码器（VAE）潜在空间构建，该空间优化像素重建，但对视觉偏移敏感；而基于语义潜在空间的模型虽更鲁棒，却无法利用大规模VAE空间预训练的视频生成模型。本文旨在克服这一两难困境。

Result: 在分布外泛化仿真基准和真实机器人实验中，Robust-WAM在不牺牲分布内性能的情况下，持续提升了多个WAM基线的成功率。

Insight: 创新点在于提出了一种通用的后训练方法，通过可学习的查询令牌将未来场景语义引入动作流，并与未来真实帧的语义前瞻对齐，同时利用动作令牌的位置编码建立时间对应关系，从而在保持大规模预训练优势的同时增强模型对视觉变化的鲁棒性。

Abstract: Mainstream World-Action Models (WAMs) adapt pretrained video generation models (VGMs) for robot control, transferring their learned dynamics prior for action prediction. These VGMs are typically trained in a variational autoencoder (VAE) latent space. However, the VAE latent space is optimized for pixel reconstruction, which rewards fine appearance detail and leaves the action prediction fragile under visual shifts. Recent works build WAMs in semantic latent space, which are more robust to appearance shifts. However, these models cannot leverage the large-scale VGM pretraining that exists only in VAE space. To overcome this dilemma, we propose Robust-WAM, a general post-training method for video-generation-based WAMs that preserves the VAE-based generative path and adds a lightweight semantic foresight alignment objective on the action stream. This retains the large-scale VGM pretraining while grounding actions in appearance-invariant dynamics that stay reliable under illumination shifts and other visual out-of-distribution conditions. Specifically, we employ learnable query tokens to bring future-scene semantics into the action stream by aligning their output hidden states with the semantic foresight of future ground-truth frames. To establish the temporal correspondence between each query and the future step it describes, we give it the positional encoding of the matching action tokens. Experiments on out-of-distribution generalization simulation benchmarks and a real-robot setup show that our Robust-WAM consistently improves the success rates of multiple WAM baselines without sacrificing in-distribution performance.

[73] Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models cs.CVPDF

Jingyan Jiang, Yaru Sun, Xiao Chen, Jiazhen Huang, Caiting Li

TL;DR: 本文针对视觉语言模型在测试时自适应（TTA）过程中校准性能下降的问题，提出了一种名为零样本锚定熵校准（ZAEC）的无标签后处理方法。该方法利用零样本熵作为样本特定的不确定性参考，通过最小化温度缩放选择性恢复被锐化预测的熵，从而改善预测置信度的可靠性，同时保持分类准确性和类别排序。

Details

Motivation: 测试时自适应（TTA）虽然能提升视觉语言模型在分布偏移下的识别准确率，但往往会降低校准性能，导致预测置信度不可靠，影响下游决策。现有无标签校准方法要么与提示优化耦合，要么依赖仅能粗略表征预测分布的logit范围统计量，存在局限性。

Result: 在五种TTA方法和15个数据集上的实验表明，ZAEC在ViT-B/16模型上实现了最低的后处理宏观平均期望校准误差（ECE），并在RN50模型上取得了一致的性能提升，同时保持了分类准确性。

Insight: 论文的创新点在于揭示了TTA可能导致‘预测保持锐化’的失效模式（即置信度增加而熵减少，但预测正确性不变），并据此提出以零样本熵作为不确定性锚点的校准策略。从客观角度看，该方法无需标注校准数据或学习参数，是一种轻量且通用的后处理方案，有效平衡了准确性与校准性能。

Abstract: Test-time adaptation (TTA) can improve the recognition accuracy of vision-language models under distribution shift, but often degrades calibration, making predictive confidence unreliable for downstream decision-making. Many existing label-free calibration approaches are either coupled to prompt optimization or rely on logit-range statistics that provide only a coarse characterization of the predictive distribution. We show that TTA can increase confidence and reduce entropy even when the top-1 prediction and its correctness remain unchanged, a failure mode we term prediction-preserving sharpening. Across diverse TTA methods and benchmarks, larger entropy reductions relative to paired zero-shot predictions are associated with greater increases in Expected Calibration Error (ECE). On entropy-reduced samples, confidence gains also tend to exceed accuracy gains. Based on these findings, we propose Zero-Shot-Anchored Entropy Calibration (ZAEC), a label-free post-hoc method that uses zero-shot entropy as a sample-specific uncertainty reference. ZAEC selectively restores the zero-shot entropy of sharpened predictions through minimal temperature scaling while leaving all other predictions unchanged. It requires no labeled calibration data or learned parameters and preserves class rankings and classification accuracy. Across five TTA methods and 15 datasets, ZAEC achieves the lowest post-hoc macro-average ECE on ViT-B/16, with consistent gains on RN50.

[74] Big, Bright, or Invisible: A Frozen-Feature Benchmark of 3D CT Foundation Models cs.CV | cs.AIPDF

Maulik Chevli, Johannes Brandt, Rickmer Braren, Daniel Rueckert, Philip Müller

TL;DR: 本研究对十个冻结的3D CT基础模型编码器进行了系统性评估，使用k近邻、零样本提示和线性探测等方法，在三个胸部CT扫描队列（包括一个未见过的内部临床数据集）上测试其诊断广度。研究发现，没有通用的最先进模型，模型排名因评估环境而异；性能主要取决于病变的物理特性（对比度和空间范围），而非模型架构。

Details

Motivation: 常规CT解读是全面的，需要捕捉整个扫描体积中的偶发发现。本研究旨在评估3D CT基础模型在提供解剖和病理的通用表示方面的诊断广度，以协助这一过程。

Result: 评估结果显示，结合细粒度图像标记化和视觉语言对齐的模型通常表现最佳，但一个轻量级的有监督编码器仍极具竞争力。在受控的器官内比较中，广泛或高对比度的异常（如设备和积液）能被可靠地检测，而小、低对比度的局灶性病变对所有评估的编码器都构成持续挑战。

Insight: 论文的创新点在于通过系统性基准测试揭示了3D CT基础模型性能的关键瓶颈是物理因素（病变对比度和大小），而非模型架构本身。客观分析认为，其核心洞察是全局池化嵌入的固有局限性，并指出准确表示小、低对比度结构需要区域或病灶级别的预训练。

Abstract: Routine CT interpretation is inherently comprehensive, capturing incidental findings across the entire scan volume. 3D CT foundation models could assist this process by providing generalizable representations of anatomy and pathology. To evaluate their diagnostic breadth, we benchmark ten frozen CT encoders across three cohorts of thoracic CT scans, including an unseen internal clinical dataset, using $k$-nearest neighbors, zero-shot prompting, and linear probing. We find no universal state-of-the-art, with rankings fluctuating significantly depending on the evaluation context. While models combining fine-grained image tokenization with vision-language alignment generally perform best, a lightweight supervised encoder remains highly competitive, demonstrating that explicit labels can effectively substitute for scale. Crucially, rather than model architecture, we observe that the primary determinant of performance is a physical bottleneck: a finding’s detectability scales with its contrast against surrounding tissue and its spatial extent. Through controlled within-organ comparisons, we empirically demonstrate that widespread or high-contrast abnormalities, such as devices and effusions, are reliably recovered. Conversely, small, low-contrast focal lesions remain a persistent challenge across all evaluated encoders. We attribute this to the inherent limitations of globally pooled embeddings, suggesting that accurately representing small, low-contrast structures will require region- or lesion-level pretraining.

[75] Diff-VF: Training-free High-quality Long Video Generation via Diffusion Model cs.CVPDF

Haoning Yang, Xinyuan Chen, Yaohui Wang, Guo Lu

TL;DR: 本文提出Diff-VF，一个无需训练、即插即用且模型无关的框架，旨在将现有的短视频扩散模型转换为高质量长视频生成器。该框架通过混合噪声初始化、加权窗口采样和时间扩展采样三种策略，有效提升长视频的时序一致性和运动多样性，并进一步通过跳跃残差引导实现长视频增强。

Details

Motivation: 现有视频扩散模型大多在短视频上训练，直接外推生成长视频时，难以保持长程时序一致性且运动多样性下降。本文旨在解决长视频生成中的一致性与动态性平衡问题。

Result: 在VBench-Long基准测试中，Diff-VF在时序一致性和运动多样性之间取得了比基础模型及FreeNoise、FreeLong、RIFLEx等无需训练基线更优的平衡，同时保持了有竞争力的逐帧质量。实验表明该方法适用于具有不同时空建模策略的视频扩散模型。

Insight: 创新点在于提出了一种无需微调基础模型的训练外长视频生成框架，通过噪声初始化、采样策略和时序融合的协同设计来约束全局语义、消除窗口间不连续并建立长程依赖。可借鉴之处包括其模型无关的即插即用架构以及时间步相关的融合与引导机制。

Abstract: Recently, diffusion models have made great progress in video generation. However, most existing video diffusion models are trained with short videos, and degrade when extrapolated to long videos, struggling to maintain long-range temporal coherence while retaining diverse motions. To generate consistent, high-quality and dynamic long videos, we propose Diff-VF, a training-free, plug-and-play and model-agnostic framework that converts existing short-video diffusion backbones into long-video generators without modifying or fine-tuning the base model. Diff-VF couples three complementary strategies: Hybrid Noise Initialization (HNI) to constrain global semantics, Weighted Window Sampling (WWS) to remove inter-window discontinuities, and Temporal Extended Sampling (TES) to establish long-range dependencies with a timestep-varying fusion. We further extend Diff-VF to long-video enhancement via Skip Residual Guidance that balances fidelity and realism through timestep-dependent guidance. VBench-Long evaluation results show that Diff-VF achieves a more favorable balance between temporal coherence and motion diversity than base models and recent training-free long video generation baselines, including FreeNoise, FreeLong, and RIFLEx, while maintaining competitive frame-wise quality. Experiments on two base models demonstrate the applicability to video diffusion models with different spatial-temporal modeling strategies. Extensive ablations validate the contribution of each component and hyperparameters.

[76] Multi-Year Geospatial Reasoning using Interannually-Consistent Historical Predictions as a Free Input Modality cs.CVPDF

Syed Roshaan Ali Shah, Kasper Bonte, David Bekaert, Kristof Van Tricht, Dieter Wens

TL;DR: 该论文提出了一种利用历史预测作为额外输入模态的深度学习方法，用于提升多年度地理空间推理任务的性能。通过将系统自身积累的过去预测档案和外部提供的辅助掩码整合到模型中，将单年度像素分类器转变为跨年度推理模型。

Details

Motivation: 现有地球观测产品生成系统通常每年使用相同模型处理新数据，但未充分利用积累的历史预测信号和外部辅助层，这些信号通常仅用于基于规则的后处理或固定输入掩码。

Result: 在欧洲约540万标记像素数据集上，添加历史预测使作物类别F1分数提升1.6个百分点，显著改善了召回率偏差问题，其中多年生作物和树木作物提升最大（橄榄+4.6、水果+3.7、坚果+3.2个百分点）。

Insight: 创新点在于提出作物类型嵌入编码器，将历史预测表示为置信度加权的时间有序分类标记，并引入跨年度注意力机制；同时系统研究了外部植被掩码在模型输入输出中的表示方法，为任何生成分类图的循环地理空间或基础模型提供了低成本改进方案。

Abstract: Machine learning, and deep networks in particular, are increasingly used to derive higher-level Earth observation (EO) products such as annual land-cover and crop-type maps. Many are generated operationally: each year a new acquisition is processed, typically with the same model, extending a multi-year archive. In the process these systems accumulate two kinds of useful signal that are almost never fed back into the model: the system’s own archive of past predictions, and ancillary layers produced by other partners in a processing consortium. Both are normally used outside the network, as rule-based post-processing or a fixed input mask. Using the Copernicus Land Monitoring Service High Resolution Layer (HRL) Croplands crop-type product as a testbed, we show that bringing both signals inside the model turns a single-year, single-task pixel classifier into one that reasons across years. We introduce a Crop Type (CTY) embedding encoder that represents each past prediction as a confidence-scaled, time-ordered categorical token and attends over the year axis, and we study how the externally provided Base Vegetation Layer (BVL) mask should be represented in the model’s inputs and outputs. To compare designs fairly when they relabel non-crop pixels, we evaluate on the 18 crop classes only and report precision and recall separately. On a pan-European dataset of about 5.4M labelled pixels, adding the prediction history raises crop-only F1 by 1.6 percentage points (pp) and, more importantly, corrects a recall-skewed error profile, with the largest gains on perennial and tree crops (olives +4.6, fruits +3.7, nuts +3.2 pp). Representing the BVL mask consistently in both the history and the target year adds about 2.5 pp on the crop classes. The approach is a low-cost recipe for any recurring geospatial or foundation model that emits class maps.

[77] Universal Concept Disruption for SAM3 Image Segmentation cs.CVPDF

Hao Wang, Yuxuan Zhang, Wei Yang

TL;DR: 本文提出了首个针对SAM3图像分割模型的通用跨概念对抗攻击方法UCD，通过从（图像，名词短语）对中学习单一有界图像扰动，联合破坏文本条件输入路径、最大化提示共享视觉特征的差异、抑制最终存在门控概念分数，并通过面积塌缩和干净掩码Dice破坏来腐蚀保留掩码的空间有效性。

Details

Motivation: SAM3将可提示分割从几何驱动的掩码预测扩展到开放词汇概念分割，但其存在门控设计在对抗鲁棒性方面尚未被探索，本文旨在研究并攻击SAM3作为集成概念-接地系统的脆弱性。

Result: 在SACo-Gold、LVIS、RefCOCO、PhraseCut和OpenImages数据集上，UCD在匹配评估协议下始终优于所有基线方法，将平均掩码AP从59.43降至18.73，平均cgF1从50.32降至20.49，且学习到的扰动无需重新优化即可迁移到SAM3.1和SAM3视频推理。

Insight: 创新点在于首次针对SAM3的集成概念-接地系统设计通用跨概念对抗攻击，通过联合多路径破坏策略（包括文本条件输入、视觉特征、概念分数和空间掩码有效性）实现高效攻击，并展示了其跨模型和跨任务（视频）的迁移性，揭示了SAM3类模型在对抗鲁棒性方面的潜在弱点。

Abstract: SAM3 extends promptable segmentation from geometry-driven mask prediction to open-vocabulary concept segmentation, where a text-conditioned grounding model decides whether a concept is present and segments all matching instances. While this presence-gated design improves concept-level prediction, its adversarial robustness remains unexplored. In this paper, we introduce Universal Concept Disruption (UCD), the first universal cross-concept adversarial attack tailored to SAM3 image segmentation. UCD learns a single bounded image perturbation from (image, noun-phrase) pairs and attacks SAM3 as an integrated concept-grounding system. It jointly disrupts the text-conditioned input path, maximizes divergence in prompt-shared visual features, suppresses the final presence-gated concept scores, and corrupts the spatial validity of retained masks through area collapse and clean-mask Dice disruption. Across SACo-Gold, LVIS, RefCOCO, PhraseCut, and OpenImages datasets, UCD consistently outperforms all baselines under a matched evaluation protocol, reducing average mask AP from 59.43 to 18.73 and average cgF1 from 50.32 to 20.49. The learned perturbation also transfers to SAM3.1 and to SAM3 video inference without re-optimization, while prompt ensembling, lightweight head fine-tuning, and temporal filtering provide limited recovery.

[78] PaCoNet: Deep Data Extraction for Parallel Coordinates cs.CVPDF

Poonam Poonam, Hannah Kniesel, Pere-Pau Vázquez, Timo Ropinski

TL;DR: 本文提出了PaCoNet，这是首个专门用于从平行坐标图中提取数据的方法。该方法不仅能提取线条坐标，还能提取单个数据样本以进行进一步分析。作者还创建了一个大规模平行坐标数据集用于训练和测试。

Details

Motivation: 当前的数据提取研究主要集中在条形图、折线图等低维可视化图表，而平行坐标作为一种广泛使用的高维数据可视化方法，其自动数据提取问题尚未得到充分探索。由于设计不佳或数据密集时平行坐标图会变得杂乱难懂，因此其自动化数据提取具有重要价值。

Result: PaCoNet是首个为平行坐标分析定制的深度学习方法，实验表明其性能显著优于未经调整的基线模型。

Insight: 主要创新点在于首次提出了针对平行坐标图的深度学习数据提取方法，并构建了相应的大规模数据集，为复杂可视化分析奠定了基础，推动了计算机视觉与数据可视化的交叉领域发展。

Abstract: Extracting data from visualizations has long challenged computer vision, with current research focused on bar, line, and pie charts, among other low-dimensional visualizations. However, parallel coordinates as a widely used high-dimensional data visualization approach, remain largely unexplored in this context. As parallel coordinate plots can quickly become cluttered and difficult to interpret when poorly designed or densely populated, automated data extraction from such visualizations is of particular interest. In this paper, we propose PaCoNet, the first approach for parallel coordinate data extraction. PaCoNet not only extracts line coordinates, but also enables the extraction of individual data samples for further analysis. Towards this end, we make the following contributions. We present the first deep learning approach tailored for parallel coordinate analysis, and demonstrate that it outperforms unadapted baselines by a significant margin. We further introduce a large-scale parallel coordinate dataset for training and testing. Together, these key contributions enable for the first time the automated analysis and redesign of parallel coordinate plots. PaCoNet thus lays the groundwork for complex visualization analysis, and further advances the intersection of computer vision and data visualization. All code, trained models, and data generation scripts will be made publicly available upon acceptance of the paper.

[79] Wan-Animate-2: Pushing the Application Boundaries of Character Animation cs.CVPDF

Guangyuan Wang, Li Hu, Dechao Meng, Zhongyi Zhang, Peng Zhang

TL;DR: 本文提出了Wan-Animate-2，一个端到端的角色动画框架，它通过重新设计的扩散Transformer直接处理驱动视频，消除了中间运动提取器，从而在运动保真度和身份保持上表现优异。论文还引入了文本驱动的视点控制功能，并提出了一个名为Wan-Animate-2-Lite的高效变体，通过三阶段训练范式将推理延迟降低到实时阈值，为交互式应用（如数字化身和直播主播）开启了新的部署场景。

Details

Motivation: 现有角色动画方法存在三大局限：基于显式运动表示的方法易受提取错误和身份漂移影响；基于隐式运动特征的方法会因压缩而丢失细粒度动态；上下文学习方法则计算成本过高。此外，所有现有系统都设计用于离线合成，无法满足交互式应用的实时性要求。

Result: 定性评估和用户研究表明，Wan-Animate-2在不同角色和运动模式上实现了高保真动画结果。其高效变体Wan-Animate-2-Lite通过创新训练方法，成功将推理延迟降至实时水平。

Insight: 主要创新点在于：1) 完全摒弃中间运动提取器，采用端到端的扩散Transformer架构直接处理驱动视频，提升了运动保真度和身份保持；2) 引入了文本驱动的视点控制，将输出相机视角与驱动视频解耦，这是依赖显式运动表示的先前方法很少支持的能力；3) 提出了一个包含教师强制预训练（带错误缓冲机制）和自强制蒸馏（分块反向传播）的三阶段训练范式，实现了从高质量生成到实时推理的模型高效化路径。

Abstract: Character image animation remains a foundational yet challenging task in computer vision. Existing approaches can be broadly categorized into three paradigms: methods based on explicit motion representations suffer from extraction errors and identity drift; methods based on implicit motion features lose fine-grained dynamics through compression; and in-context learning approaches avoid intermediate representations but incur prohibitive computational costs. Furthermore, all current systems are designed for offline synthesis, unable to meet the real-time requirements of interactive applications such as digital avatars and live-streaming hosts. To address these limitations, we present Wan-Animate-2, an end-to-end character animation framework that directly consumes the driving video within a redesigned Diffusion Transformer. Our architecture achieves superior motion fidelity and identity preservation by eliminating intermediate motion extractors entirely. We further introduce text driven viewpoint control that decouples the output camera perspective from the driving video–a capability rarely supported by prior character animation methods that rely on explicit motion representations. Beyond generation quality, we present Wan-Animate-2-Lite, an efficient variant that reduces inference latency to real-time thresholds through a three-stage training paradigm: teacher forcing pretraining with error buffer mechanism, and Self-Forcing distillation with chunk-wise backpropagation. This enables streaming character animation for interactive applications, opening new deployment scenarios that were previously infeasible. Qualitative evaluations and user studies demonstrate that Wan-Animate-2 achieves high-fidelity animation results across diverse characters and motion patterns. To foster further research and community development, we will release the Wan-Animate-2-Base model weights to the public.

[80] Learning from Failures: Retrieval-Centric CoT via Hard Negatives for Unified Multimodal Retrieval cs.CVPDF

Zelong Sun, Jun Wang, Kaicheng Yang, Tiancheng Gu, Ziyong Feng

TL;DR: 本文提出了一种名为UniME-R1的检索中心思维链框架，用于统一多模态检索。该框架通过分析初始检索结果中的困难负样本来生成针对检索的思维链，从而纠正嵌入器的误解，并通过重排序或重新检索来提升性能。

Details

Motivation: 现有基于大视觉语言模型的检索器直接编码原始多模态输入时，容易忽略细粒度判别性线索，导致对语义相似候选的混淆。而现有的思维链方法仅从查询本身生成推理，未能针对检索器的具体误解进行优化。

Result: 在MMEB-V2和一系列通用多模态检索基准测试上的广泛实验表明，UniME-R1持续超越了强基线模型，提升了检索性能。

Insight: 核心创新在于提出了检索中心思维链的概念，将推理过程与检索反馈（特别是困难负样本）相结合。方法上的创新点包括嵌入器-顾问框架、利用困难负样本模拟检索失败进行训练，以及结合监督学习和面向检索的强化学习来对齐顾问与检索结果。

Abstract: Unified multimodal retrieval aims to identify candidates that satisfy complex user intent expressed through heterogeneous inputs. Although Large Vision-Language Model (LVLM)-based retrievers are efficient and scalable, directly encoding raw multimodal inputs often misses fine-grained discriminative cues, leading to confusion among semantically similar candidates. Recent methods mitigate this limitation by generating Chain-of-Thought (CoT) rationales to enrich the query representation. However, such reasoning is typically derived from the query alone: it explains what the query describes, but not what the retriever misunderstands. We argue that effective retrieval reasoning should instead be conditioned on retrieval feedback. Based on this insight, we introduce UniME-R1, an embedder-adviser framework that learns to reason over initially retrieved candidates and generate Retrieval-Centric Chain-of-Thought (RC-CoT). The adviser analyzes candidates individually to identify the discriminative cues confused by the embedder. If the target appears in the initial top-k set, UniME-R1 directly reranks the candidates; otherwise, it generates RC-CoT to refine the retrieval direction and performs full-corpus re-retrieval with a dual-mode embedder. To train the framework, we mine hard negatives to simulate realistic retrieval failures, jointly optimize direct retrieval and RC-CoT-augmented retrieval, and align the adviser with retrieval outcomes through supervised learning and retrieval-oriented reinforcement learning. Extensive experiments on MMEB-V2 and a diverse set of general multimodal retrieval benchmarks demonstrate that UniME-R1 consistently improves retrieval performance over strong baselines.

[81] The Next Screenshot Knows: Gated Hindsight Distillation for Mobile GUI Agents cs.CVPDF

Weiwei Li, Junzhuo Liu, Tong Chu, Hengfu Yu, Wen Li

TL;DR: 本文提出了一种名为Gated Hindsight Distillation (GHD)的新方法，用于训练移动GUI智能体。该方法在训练时利用下一张屏幕截图作为特权信息，通过一个参数共享的教师模型来重新评估学生模型在可观察轨迹前缀上的策略响应，从而帮助智能体理解动作正确性的证据。

Details

Motivation: 标准离线训练方法将成功交互轨迹分解为前缀-动作对，丢弃后续观察，这导致模型无法学习动作正确性的关键证据（这些证据往往只在后续屏幕上出现），从而限制了模仿学习的效果。

Result: 在AndroidWorld和AndroidLab基准测试中，GHD方法相比GRPO（Group Relative Policy Optimization）提高了任务成功率。

Insight: 核心创新点在于利用‘后见之明’（hindsight）信息（即下一屏幕截图）作为训练时的特权监督信号，并通过门控机制仅在学生模型失败且教师模型能恢复演示动作时才进行知识蒸馏，这有效引导了模型学习动作背后的推理依据。

Abstract: GUI agents are commonly trained offline from successful interaction trajectories. Standard training decomposes each trajectory into prefix-action pairs: the agent predicts an action from the current screen and interaction history, while the subsequent observation is discarded. This removes the rationale of why an action is correct: the evidence often appears only on the subsequent screen. For example, to enable Soft Wrap, the agent should click Edit or View, but nothing reveals this until the menu opens. Without such evidence, standard imitation gives the model little chance of ever sampling and thus learning the correct reasoning. To address this issue, we propose Gated Hindsight Distillation (GHD), which uses the next screenshot as privileged information during training. A student predicts from the observable trajectory prefix, while a parameter-sharing teacher additionally observes the next screenshot and re-scores the student’s on-policy responses. We apply distillation only when the student fails and the hindsight-conditioned teacher recovers the demonstrated action. GHD improves task success over GRPO on AndroidWorld and AndroidLab across two vision-language models. The code and checkpoints will be made available.

[82] Domain-Grounded Candidate Selection for Agentic Image Editing: A Shadow Removal Case cs.CV | cs.AIPDF

Shilin Hu, Jingyi Xu, Dimitris Samaras, Hieu Le

TL;DR: 本文研究了商用视觉语言模型在低层视觉任务（如阴影去除）中的应用潜力，发现直接使用生成式编辑器虽能产生干净的阴影去除结果，但存在内容再生、物体幻觉等物理错误。为此，作者提出了一种基于物理先验的代理候选选择流程，通过引导生成、评估筛选和候选选择来平衡阴影去除与场景保持，在ShadowRemovalRefine基准上显著降低了CDD指标。

Details

Motivation: 探讨商用视觉语言模型是否能够替代传统的基于物理先验的低层视觉方法，特别是在阴影去除这种受场景几何、光照、材质等多因素影响且配对数据稀缺的任务中。

Result: 在ShadowRemovalRefine基准测试中，所提出的基于物理导向的流程取得了0.0075的CDD值，比之前最强方法降低了至少47%，达到了新的SOTA水平。

Insight: 创新点在于将物理先验（如将阴影视为光照遮挡效应而非材质或物体结构）融入生成式模型的引导与评估流程中，通过代理候选选择机制约束生成过程，提升了结果的物理合理性与一致性，证明了经典低层视觉先验在引导生成式模型中的持续价值。

Abstract: Commercial vision-language models are reshaping computer vision, with visual priors broad enough to rival task-specific systems. This raises a natural question: do they reduce the need for classic, physics-informed low-level vision? We study this through shadow removal, a problem shaped by scene geometry, illumination, materials, and occluders, where paired shadow and shadow-free data are hard to collect at scale. We find that a commercial generative editor, used directly, can produce clean shadow-free edits that preserve surface texture and local appearance. However, this comes with a new failure mode: the same editor can regenerate scene content, hallucinate objects, or misread a shadow as material or geometry, producing plausible but physically wrong edits. We address this with an agentic candidate-selection pipeline: the editor generates a guided probe, an evaluator screens for major failures, retries when needed, samples multiple candidates, filters them, and selects a final result balancing shadow removal against scene preservation. Grounding this process in shadow-formation physics makes it more reliable: prompting the generator and evaluator to treat shadows as illumination effects caused by light occlusion, not material or object structure, measurably improves quality and consistency. On the ShadowRemovalRefine benchmark, our physics-oriented pipeline achieves a CDD of 0.0075, reducing CDD by at least 47% over the strongest prior method. These results suggest that commercial vision-language models do not replace classic low-level vision priors; instead, such priors remain useful for constraining and steering physically underconstrained generation.

[83] BendTwin: Robust Dense-to-Sparse Physical Reconstruction with Bending-Aware Differentiable Spring-Mass Models cs.CVPDF

Yixiong Jing, Qi Wang, Lin Chen, Junwei Jiang, Guangming Wang

TL;DR: 本文提出了BendTwin，一种用于从视频重建和预测可变形物体的、具有弯曲感知能力的可微分弹簧-质点框架。该方法通过在局部表面三元组上引入弯曲刚度和阻尼，惩罚偏离静止角度的变形，从而改善了仅使用轴向弹簧的传统方法的机械稳定性。

Details

Motivation: 现有基于弹簧-质点的物理驱动重建方法通常仅依赖轴向弹簧，这过度简化了结构力学，且在物理图被粗化时可能变得机械欠约束，限制了其保持稳定局部变形的能力。

Result: 实验表明，BendTwin在性能上持续优于仅使用轴向弹簧的PhysTwin基线。消融研究进一步证明，弯曲约束能在不同下采样率下保持系统稳定性，并持续改进原始的PhysTwin公式。

Insight: 核心创新点在于引入了针对局部表面三元组的弯曲刚度和阻尼约束，这正则化了高阶变形，提高了机械稳定性，同时保持了弹簧-质点系统的简洁性，为从稀疏视角RGB-D视频构建机械上可信的数字孪生提供了有效方法。

Abstract: Reconstructing objects with mechanical properties from video observations enables physically consistent dynamic prediction, benefiting robotics planning and interaction. Existing spring–mass based physical driven reconstruction approaches offer efficient and differentiable physical reconstruction, but they typically rely on axial springs alone. Such formulations oversimplify the underlying structural mechanics and can become mechanically under-constrained when the physical graph is coarsened, limiting their ability to preserve stable local deformation. We present BendTwin, a bending-aware differentiable spring–mass framework for video-based reconstruction and future prediction of deformable objects. BendTwin introduces bending stiffness and damping over local surface triplets, penalizing deviations from rest angles and regularizing higher-order deformation. These bending constraints improve mechanical stability while preserving the simplicity of spring–mass system. Experiments show that BendTwin consistently outperforms the axial-only PhysTwin baseline. Ablation studies further demonstrate that the bending constraints maintain system stability across different downsampling ratios and consistently improve upon the original PhysTwin formulation. Overall, BendTwin provides an effective approach for constructing mechanically faithful digital twins from sparse-view RGB-D videos.

[84] Sample-Adaptive Latent Rewards for Uncertainty-Guided Diffusion Post-Training cs.CVPDF

Rui Li, Yuanzhi Liang, Ke Hao, Ziqiao Weng, Haibin Huang

TL;DR: 本文提出了SURE框架，一个用于图像和视频扩散模型的后训练统一潜在空间框架。该框架包含一个样本自适应潜在奖励模型（SURE-LRM），用于预测每个噪声潜在向量的高斯效用分布（均值为奖励分数，方差为不确定性），以及一个不确定性引导的奖励反馈学习机制（SURE-REFL），利用预测的不确定性来指导密集的后训练过程。

Details

Motivation: 现有潜在奖励模型仅输出标量分数，无法估计每个预测的不确定性，导致生成器无法判断反馈的可靠性，可能引发错误的优化方向（如奖励破解）。本文旨在解决这一问题，使扩散模型的后训练对齐过程更高效、更稳定。

Result: 实验表明，SURE-LRM在偏好预测上优于强基线模型。SURE-REFL在多项指标上达到了最先进的性能，并进一步提升了优化稳定性，在VBench评估中获得了最高的质量、语义和总分。

Insight: 核心创新在于将潜在奖励建模为高斯分布，从而同时输出奖励分数和不确定性估计，并利用不确定性作为权重来引导后训练过程，实现了不确定性引导的密集反馈。整个过程完全在潜在空间中进行，无需像素空间解码或完整的去噪图，提高了效率。

Abstract: Latent reward models can supervise visual diffusion models without decoding intermediate states into pixel space. This makes alignment with human preferences more efficient. However, existing latent reward models output only scalar scores. They do not estimate the uncertainty of each prediction. The generator therefore cannot determine which feedback is reliable. This can drive optimization in the wrong direction and lead to reward hacking. We propose \textsc{SURE}, a unified latent-space framework for image and video diffusion models. It learns reward distributions and directly uses their reliability to guide dense post-training. First, we propose sample-adaptive latent reward model (\textsc{SURE-LRM}). It predicts a Gaussian utility for each noisy latent. Its mean predicts the reward score. Its variance reflect the uncertainty of prediction without human annotation. The learned distribution then guides post-training through uncertainty-guided reward feedback learning (\textsc{SURE-REFL}). This method provides uncertainty-guided dense feedback along the denoising trajectory. At selected transitions, \textsc{SURE-REFL} queries the frozen \textsc{SURE-LRM}. It converts detached variance into reliability weights for samples at the same transition. Each weighted reward is backpropagated only through its local transition. The entire process remains in latent space and requires neither pixel-space decoding nor the full denoising graph. Experiments show that \textsc{SURE-LRM} improves preference prediction over strong baselines. \textsc{SURE-REFL} achieves the sota performance among various metrics and further improves optimization stability. It also achieves the highest VBench quality, semantic, and total scores among the evaluated methods.

[85] Depth-Guided Video Object Counting in Crowded Scenes cs.CV | cs.AIPDF

Yuanjing Xu, Xinyan Liu, Weidong Chen, Zixuan Zou, Linhao Zhang

TL;DR: 本文提出了一种深度引导的视频目标计数方法，旨在通过整合深度信息来提升拥挤场景下目标计数的鲁棒性。该方法包括一个深度引导检测器（DG-Det）和一个通用的后处理流程，通过多尺度RGB-D交叉注意力和显式遮挡预测来增强空间理解，并引入统一的去重框架以消除跨帧冗余计数。

Details

Motivation: 现有视频目标计数方法主要依赖RGB信息，在拥挤和遮挡场景下的区分能力有限，因此需要引入深度信息来提升性能。

Result: 在提出的新RGB-D视频目标计数数据集上，该方法相比现有基线在MAE指标上降低了62.01%，并在RMSE上取得了一致的改进。

Insight: 创新点在于首次将深度信息整合到视频目标计数中，通过RGB-D交叉注意力和显式遮挡建模来应对拥挤遮挡挑战，并发布了首个包含深度信息的多类别RGB-D视频计数数据集以推动未来研究。

Abstract: Our primary objective is to advance video object counting in crowded scenes, aiming to robustly count all instances of a target category based on given text or visual prompts. Existing methods rely on RGB information, limiting their discriminative ability in crowded and occluded conditions. To address this, we propose a Depth-Guided Detector (DG-Det) along with a general post-processing pipeline. By integrating depth cues with multi-scale RGB-D cross-attention and explicit occlusion prediction, our method enhances spatial understanding and achieves robust detection in crowded and occluded scenes. Furthermore, we introduce a unified de-duplication framework to eliminate cross-frame redundant counting. To facilitate future research, we also release a new RGB-D Video Object Counting dataset featuring depth information and multiple object categories persequence. Extensive experiments demonstrate that our method achieves a 62.01% reduction in MAE compared to existing baselines, and also produces consistent improvements in RMSE. We provide the source code at https://github.com/streamer-AP/DG-Net and the dataset at https://huggingface.co/datasets/aerospace123/RGBD-VideoCount.

[86] HOPE: Hand-Object Pressure Estimation from Monocular Videos cs.CVPDF

Subin Jeon, Byungjun Kim, Hanbyul Joo

TL;DR: 该论文提出了HOPE框架，用于从单目视频中估计手与物体交互时的压力。该方法将压力估计建模为以手为中心的视频预测问题，预测手网格顶点上随时间演化的法向压力和接触状态，从而形成一个独立于物体形状和传感器布局的统一输出空间。

Details

Motivation: 现有的基于视觉的压力估计方法主要局限于平面表面和单张图像输入，难以应用于具有多样物体的动态手-物交互场景。

Result: 在OpenTouch、PressureVisionDB和手-物接触基准测试上的实验验证了HOPE在物体压力、表面压力和接触监督的手-物交互设置中的有效性。尽管主要使用戴手套手视频的度量压力监督，HOPE能泛化到裸手的第一人称和野外视频，其联合接触和压力预测能力超越了仅接触或平面压力基线方法。

Insight: 创新点在于将压力估计重新定义为手中心的视频预测问题，并提出了一个统一的框架，通过将不同来源的触觉数据（如触觉手套压力、平面传感器压力）提升到共享的手顶点空间来整合监督信号，以及使用一个以顶点为锚点的视频Transformer和接触门控压力头来增强预测的物理一致性。

Abstract: Estimating physical pressure from vision is essential for understanding contact-rich hand-object interaction. However, prior vision-based pressure estimation methods are largely limited to planar surfaces and single image input, making them difficult to apply to dynamic hand-object interaction with diverse objects. We instead formulate pressure estimation as a hand-centric video prediction problem with monocular video as input. This formulation predicts temporally evolving per-vertex normal pressure and contact directly on the hand mesh, yielding a unified output space independent of object shape and sensor layout. Building on this formulation, we propose \textbf{HOPE}, a framework with two key components. First, we lift tactile-glove pressure, planar-sensor pressure, and distance-based hand-object contact annotations into a shared hand vertex space, allowing bare-hand contact data to regularize pressure learning where metric labels are unavailable. Second, we introduce a vertex-anchored video transformer that treats each vertex as a persistent token, aggregates visual features and hand pose over time, and uses a contact-gated pressure head to enforce that pressure vanishes without contact. Experiments on OpenTouch, PressureVisionDB, and hand-object contact benchmarks validate HOPE across object-pressure, surface-pressure, and contact-supervised HOI settings. Despite using metric pressure supervision primarily from gloved-hand videos, HOPE generalizes to bare-hand egocentric and in-the-wild videos, producing joint contact and pressure predictions beyond the scope of contact-only or planar-pressure baselines.

[87] EmoWorld: A Decoupled Affective Field for Controllable Emotional Video Generation cs.CVPDF

Bingyuan Wang, Baistan Zhyldyzbekov, Kunyu Feng, Zeyu Wang

TL;DR: 本文提出了EmoWorld框架，用于可控的情感视频生成。该框架在冻结的视频扩散Transformer中，将全局氛围、承载情感的语义线索和时间进程解耦。通过视觉氛围引导、语义情感引导和时间情感引导三个模块，分别控制视频生成的不同情感维度。

Details

Motivation: 现有视频生成器将全局氛围、情感语义线索和时间进程耦合在单一文本条件中，限制了情感控制的精细度。EmoWorld旨在将这些因素解耦，以实现更精确、可控的情感视频生成。

Result: 在Wan2.2基准上，视觉氛围引导将目标情感对齐度提升19%，时间波动代理降低48%；语义情感引导将目标情感对齐度提升37%，检测到的情感线索增加36%；时间情感引导将过渡单调性提升15%。该框架在27种情感类别和多种生成设置下进行了评估，并支持跨多个视频DiT骨干的移植。

Insight: 创新点在于提出了一种解耦的情感控制框架，通过一次性提取可重用的情感方向和线索库，实现了对视频氛围、语义和时间三个维度的独立、可扩展的引导，且无需更新生成器参数，具有良好的可移植性和组合性。

Abstract: Emotion shapes how viewers interpret a scene, yet existing video generators entangle global atmosphere, affect-bearing semantic cues, and temporal progression within a single text condition. We present EmoWorld, a framework that decouples these factors within a frozen flow-matching video diffusion transformer (Video DiT). A one-time preparation stage extracts layer-specific affect directions and a reusable cue library from geometry-preserving neutral and emotion-edited panoramas. At inference, Visual Atmosphere Steering (VAS) injects atmosphere directions into hidden states, Semantic Affective Steering (SAS) isolates a separately scalable prompt residual for semantic cues, and Temporal Affective Steering (TAS) interpolates endpoint residual fields across denoising and video time. On Wan2.2, VAS improves target-emotion alignment by 19% while reducing a temporal-fluctuation proxy by 48%; SAS improves target-emotion alignment by 37% and increases detected affect-bearing cues by 36%; and TAS improves transition monotonicity by 15% over the strongest baseline. EmoWorld is evaluated across 27 emotion categories in text-to-video and image-to-video settings, demonstrates portability across multiple Video-DiT backbones, and supports camera-conditioned composition without updating generator parameters.

[88] Support Operation Factorization: Compositional Readout of Frozen Vision Encoders under Controlled Interventions cs.CVPDF

Zhongyao Wang, Wanli Ouyang, Taoyong Cui, Pheng Ann Heng

TL;DR: 本文提出了一种名为支持操作分解（SO-OPF）的读出方法，用于对冻结视觉编码器进行组合分析，旨在同时确定变化的内容和位置。该方法通过将单元能量分解为支持显著性和竞争性操作后验，解决了标准因子探针中存在的操作洗钱问题。在Shapes3D-Extended和COCO数据集上验证了其有效性，并揭示了不同渲染器（如DINOv3和SigLIP2）在从扁平标签学习分配时的性能边界。

Details

Motivation: 标准因子探针在分析冻结视觉编码器的组合性时，分别评估’什么变化’和’哪里变化’，可能导致多个操作重用同一预测槽位，即’操作洗钱’问题。本文旨在开发一种方法，能够同时且清晰地分解这两个方面，以更准确地评估编码器的组合理解能力。

Result: 在Shapes3D-Extended数据集上，已知因子分配达到了0.874的注入准确率，在全局图像不相交的COCO上达到0.799；从扁平标签学习分配分别达到0.769和0.762。在Shapes3D上，分解载体的学习分配准确率从密集载体的0.653提升到0.841，并消除了洗钱差距。SigLIP2在COCO上复制了这种分离效果，但在重建的MuJoCo基底上，DINOv3和SigLIP2的学习分配准确率分别降至0.569和0.484，出现显著的槽位崩溃。

Insight: 创新点在于提出了SO-OPF读出方法和注入对齐的留一单元协议，将组合分析的两个核心问题（已知网格下的组合能力与从扁平标签恢复网格的能力）明确分离。客观来看，该方法不仅提升了性能，更重要的是暴露了模型在特定渲染条件下的失败边界，而非掩盖问题，为评估冻结编码器的组合性提供了更透明和诊断性的工具。

Abstract: Compositional analysis of frozen vision encoders should determine both what changed and where it changed. Standard factor probes score these axes separately, however, and can reward multiple operations that reuse the same predicted slot. We call this failure operation laundering. We introduce an injectively aligned leave-one-cell-out protocol over support x operation grids and SO-OPF, a readout that factors cell energy into support salience and a competitive operation posterior. This formulation separates two questions that aggregate scores conflate: whether the carrier composes held-out bindings when the grid is known, and whether that grid can be recovered from flat cell labels. With frozen DINOv3 features, known factorial assignment reaches 0.874 injective accuracy on Shapes3D-Extended and 0.799 on globally image-disjoint COCO; learning the assignment from flat labels reaches 0.769 and 0.762, respectively. Under matched-axis-aware supervision on Shapes3D, the factored carrier improves learned-assignment accuracy from 0.653 to 0.841 over a dense carrier and eliminates its laundering gap. SigLIP2 replicates the COCO separation. A rebuilt MuJoCo substrate exposes a boundary: learned-assignment accuracy is 0.569 with DINOv3 and 0.484 with SigLIP2, with substantial slot collapse. Thus factored readout and injective evaluation recover held-out bindings on two substrates while exposing, rather than hiding, a renderer-specific failure boundary; they do not establish universal recovery from flat labels.

[89] MASS: Multiplayer World Models with Authoritative Shared State cs.CV | cs.HCPDF

Ziqi Cai, Siqi Yang, Yimu Wang, Zixian Gao, Yunheng Liu

TL;DR: 本文提出MASS（Multiplayer world models with Authoritative Shared State），一种用于多人环境的世界模型。它受到多人游戏架构启发，将世界动态与视图渲染解耦：一个学习的逻辑引擎从联合动作推进全局权威类型化状态，作为唯一的循环记忆和同步参考；一个学习的渲染引擎从该共享状态按需为任意请求的相机生成独立且一致的视图。该方法在多人贪吃蛇基准测试中实现了优于现有多视图基线的状态准确性和更低的跨视图不一致性，并能以1,024个并发玩家推进10,000个循环步骤的预测世界。

Details

Motivation: 当前视频世界模型在多人环境中表现不佳，因为它们将世界状态与视图相关的视觉潜在变量纠缠在一起，导致计算冗余、视图不一致和可扩展性差。MASS旨在解决这一局限。

Result: 在匹配的多人贪吃蛇基准测试上，MASS相比最先进的多视图基线实现了更高的状态准确性和更低的跨视图不一致性。它能以1,024个并发玩家推进10,000个循环步骤的预测世界。

Insight: 核心创新在于借鉴多人游戏架构，明确地将世界动态（逻辑引擎）与视图渲染（渲染引擎）解耦，并引入一个学习的、权威的全局类型化状态作为唯一的循环记忆和同步参考。这为可扩展且一致的多智能体世界模拟提供了一个实用的基础，避免了手工编写状态转移函数。

Abstract: Current video world models struggle in multiplayer environments because they entangle world state with view-dependent visual latents, leading to redundant compute, view inconsistencies, and poor scalability. We propose MAS (Multiplayer world models with Authoritative Shared State) to resolve this limitation. Inspired by multiplayer game architectures, MAS disentangles world dynamics and view rendering. A learned Logic Engine advances a global, authoritative typed state from joint actions without any hand-written transition function, acting as the sole recurrent memory and synchronization reference. From this shared state, a learned Rendering Engine generates independent and consistent views for any requested camera on demand. This explicit disentangling allows MAS to achieve superior state accuracy and lower cross-view inconsistency compared to state-of-the-art multi-view baselines on a matched multiplayer Snake benchmark. It advances predicted worlds with 1,024 concurrent players for 10,000 recurrent steps. Our results show that explicit, authoritative state modeling provides a practical foundation for scalable and consistent multi-agent world simulation.

[90] PRISM: Distribution-Gated Flow Matching for Controllable Unpaired Image Translation cs.CV | cs.AIPDF

Elad Yoshai, Natan T. Shaked

TL;DR: PRISM提出了一种基于分布门控流匹配的无配对图像翻译框架，通过学习的逐特征门控机制替代全局噪声控制，实现更精细的内容保留与外观转换。该方法在初始化阶段混合真实源特征与任务匹配的噪声，并在ODE积分过程中通过门控调节传输时机，支持通过文本或检测器进行局部覆盖而无需重新训练。

Details

Motivation: 现有基于扩散模型的无配对图像翻译方法通常使用单一全局噪声或引导值控制图像保留，无法区分需要保留的内容与需要改变的外观，导致控制精度不足。

Result: 在AFHQ猫->狗、CelebA-HQ外观转换、日->夜重光照、虚拟染色和乳腺冷冻->永久组织病理学五个自然与生物医学基准测试中，PRISM在相同划分协议下取得四个数据集的最佳Inception FID和KID分数，第五个数据集结果具有竞争力；在组织病理学任务中生成最接近理想值的细胞核计数比，表明其在目标真实性与结构保留间取得良好平衡。

Insight: 创新点包括：1）通过特征到目标分布标准化距离构建空间先验的门控机制，实现特征级保留控制；2）将任务匹配的噪声（结构保留任务用AdaIN锚定，结构改变任务用部分锚定）与门控统一框架；3）推理时支持通过外部信号（如文本或检测器）局部覆盖门控，增强可控性且无需重新训练。

Abstract: Unpaired image-to-image translation must decide, per image, what to change and what to preserve without paired supervision. Many diffusion-based unpaired translators control preservation through a single global noise or guidance value applied across the image, which cannot separate content to keep from appearance to change. We present PRISM, a GAN-free flow-matching framework that replaces this global control with a learned per-feature gate. The gate’s spatial prior is derived from each source feature’s standardized distance to the target feature distribution, so features far from the target are freed while target-consistent features are preserved. The same gate controls both the initialization, which mixes the real source latent with a task-matched corruption, and the transport timing during Ordinary Differential Equation (ODE) integration. The corruption is matched to the task, content-anchored (AdaIN) for structure-preserving translation and partially anchored for structure-changing translation, and the gate can be overridden locally at inference from text or a detector without retraining, preserving important structures of the original image while still generating realistic results. We evaluate PRISM on five natural and biomedical benchmarks (AFHQ cat->dog, CelebA-HQ appearance translation, day->night relighting, virtual staining, and breast frozen->permanent histopathology). Among the evaluated methods under a shared same-split protocol, PRISM attains the best Inception FID and KID on four benchmarks and a competitive result on the fifth, and on histopathology yields the nuclei-count ratio closest to the ideal, supporting a favorable balance between target realism and structural preservation.

cs.HC [Back]

[91] A Multi-Layer System for Ultra-High-Resolution Static 360-Degree Telepresence cs.HC | cs.CVPDF

Jiapeng Chi, Gerd Bruder, Carsten Neumann, Carolina Cruz-Neira, Dirk Reiners

TL;DR: 本文提出了一种用于静态360度远程呈现的多层超高清系统，该系统结合了8K全景相机和可旋转的4K云台变焦相机。该方法构建了一个三层表示：基于图块的超高清全景背景层、动态更新层和感兴趣区域4K层。系统通过离线拼接和实时背景抠图等技术，在有限的硬件分辨率下实现了超越原生捕获的有效分辨率，并平衡了全景保真度和交互性。

Details

Motivation: 当前360度视频远程呈现受限于捕获和显示硬件的分辨率，且针对固定视点和静态场景的优化策略研究不足。本文旨在为这类场景设计一个专门的系统，以克服硬件限制，提升沉浸感。

Result: 通过与代表性视频超分辨率方法进行比较，并进行了评估感知细节和沉浸感的用户研究。结果表明，基于图块的背景细化与用户引导的更新相结合，为静态360度远程呈现提供了一种平衡全景保真度和交互性的实用方法。

Insight: 创新点在于为静态场景量身定制的三层系统架构，特别是通过离线高细节扫描拼接生成超高清背景图块，并结合实时动态更新与用户交互的感兴趣区域流，在硬件限制下有效提升了感知分辨率。这是一种将离线预处理与实时流媒体、用户交互相结合的实用化系统设计思路。

Abstract: 360-degree video telepresence offers strong immersive potential but remains constrained by the limited resolution of current capture and display hardware. Many telepresence installations feature fixed viewpoints and largely static scenes, yet optimization strategies tailored to such setups have received limited attention. We present a multi-layer, ultra-high-resolution system for static 360-degree telepresence that combines an 8K panoramic camera with a rotatable 4K pan-tilt-zoom (PTZ) camera. Our approach builds a three-layer representation: (1) a tile-based ultra-high-resolution panoramic background, generated by offline stitching high-detail 4K PTZ scans onto the base 8K panorama to achieve effective resolution beyond native capture, and represented as a set of spatial tiles; (2) a dynamic update layer that composites foreground motions from the 8K stream via real-time high-resolution background matting; and (3) a region-of-interest 4K layer that streams a real-time PTZ view of the selected region and additionally updates the corresponding background tiles over time. We evaluate the proposed system through comparisons with representative video super-resolution approaches and a user study assessing perceived detail and immersive experience. Our results indicate that tile-based background refinement, together with user-guided updates, provides a practical way to balance panoramic fidelity and interactivity in static 360-degree telepresence.

cs.LG [Back]

[92] QEvict: Recoverable Quantized KV Eviction for Attention-Drift-Robust Long-Context Decoding cs.LG | cs.CLPDF

Ayushman Garg, Akshita Gupta, Shaswata Bhattacharya, Abhishek Gupta, Sandeep Kumar

TL;DR: 本文提出QEvict，一种可恢复的量化KV缓存逐出方案，用于解决自回归大语言模型推理中KV缓存内存占用过大的问题。该方法通过三层管理（全精度保留、量化可恢复、删除）替代传统的二元保留/删除策略，允许被逐出的令牌在后续解码中因注意力漂移而重新变得重要时被恢复，从而在固定内存预算下保留更广泛的历史上下文并保持最重要区域的全精度。

Details

Motivation: 现有基于注意力得分的KV缓存逐出策略隐含了不可逆的决策假设，即一旦令牌被逐出就无法再次使用。然而，解码过程中令牌和窗口的重要性会随着生成的查询演变而漂移，导致标准逐出策略永久丢弃了那些在全缓存模型下后续可能获得大量注意力的状态。

Result: 在长上下文理解、检索和推理基准测试中，QEvict持续优于代表性的逐出和量化基线方法，减少了错过的注意力并改善了信息保留。

Insight: 核心创新在于引入了‘可恢复逐出’的概念，通过量化中间窗口作为可恢复层，并基于累积注意力分数动态更新窗口重要性，实现了缓存管理的弹性。这挑战了传统逐出策略的不可逆假设，为KV缓存压缩提供了新思路。

Abstract: Autoregressive large language model inference is increasingly constrained by the memory footprint of the Key-Value (KV) cache. A dominant line of work reduces this footprint by evicting tokens that appear unimportant under attention-derived scores. However, such policies make an implicit irreversible decision: once a token is evicted, it cannot become useful again. We show that this assumption is brittle during decoding. Token and window importance drift as generated queries evolve, causing standard eviction policies to permanently discard states that later receive substantial attention under the full-cache model. To characterize this behaviour, we introduce Future Missed Mass and Global LIR, two diagnostics that measure future attention assigned to discarded states and the reactivation of historically inactive regions. We propose QEvict, a three-tier KV-cache management scheme that replaces binary retain-or-delete eviction with recoverable eviction. QEvict maintains high-confidence windows in full precision, stores intermediate windows in a quantized recoverable tier, and deletes only the lowest-confidence windows. During decoding, cumulative attention scores update window importance and when a quantized window becomes important again, it is dequantized and promoted to the full-precision. Under a fixed memory budget, this design preserves broader historical context while retaining exact full precision for the most important regions. Across long-context understanding, retrieval, and reasoning benchmarks, QEvict consistently improves over representative eviction and quantization baselines, reducing missed attention and improving information retention

[93] Reasoning Errors Have a Region and a Direction in the Residual-Stream Trajectory of LLMs cs.LG | cs.CLPDF

Hamed Damirchi, Ignacio Meza De la Jara, Damith Ranasinghe, Yuhang Liu, Javen Shi

TL;DR: 该论文提出了一种三流检测器，用于在大型语言模型的残差流轨迹中识别推理错误。该方法结合了残差流位移、基于向量量化的粗略区域读取器以及对归一化多层状态的精细方向读取器，以在避免引入易受捷径影响信息的同时恢复足够的状态上下文来解读推理过程。

Details

Motivation: 随着语言模型越来越多地用于需要可验证推理的任务，可靠地区分正确推理与错误推理已成为重要的实际问题。现有基于轨迹的方法在层间残差流位移中寻找信号，但位移忽略了更新源自的状态，而恢复完整状态又可能重新引入易受捷径影响的信息。

Result: 在训练期间未见过的推理基准测试上，该方法比仅使用位移的现有最佳方法提高了高达12%的选择准确率，比单层探测基线提高了21%。尽管仅在推理基准上训练，该方法在事实补全和事实验证任务上也优于所有对比的检测器，表明其信号基于正确性而非特定推理类型。消融实验进一步表明运动、区域和方向提供了互补的信号。

Insight: 核心创新在于提出了一种结合运动（位移）与受限状态视图（区域和方向）的三流检测器设计，以解读状态条件化的运动，从而更好地读取推理有效性。这表明推理有效性从状态条件化的运动中读取，比仅从静态状态或去上下文化的轨迹中读取更优。

Abstract: As language models are increasingly used for tasks that require verifiable reasoning, reliably distinguishing sound reasoning from flawed reasoning has become an important practical problem. Recent trajectory-based methods seek this signal in layerwise residual-stream displacements, which capture how representations change while attenuating some stable, token-specific information. However, displacement omits the state from which an update originates, whereas restoring the full state risks reintroducing shortcut-prone information. We identify this trade-off and propose a three-stream detector that combines motion with two restricted views of location. A coarse region reader based on vector quantization and a fine direction reader over normalized multi-layer states. This design restores enough state context to interpret the motion without returning to full-state probing. On reasoning benchmarks unseen during training, our method improves selection accuracy by up to 12% over the displacement-only state of the art and 21% over single-layer probing baselines. Although trained only on reasoning benchmarks, it also reads factual completion and fact verification, ahead of every detector we compare against, which places the signal on correctness rather than on a kind of reasoning. Ablations further show that motion, region, and direction provide complementary signals. These results suggest that reasoning validity is better read from state-conditioned motion than from either static states or decontextualized trajectories alone.

[94] RRC: Unlocking Generative Reward Models in LLM Reinforcement Learning via Ranking-Based Reward Construction cs.LG | cs.CLPDF

Chenglong Wang, Ziming Zhu, Yifu Huo, Bei Li, Qiaozhi He

TL;DR: 本文提出了一种基于排序的奖励构建方法，旨在解决生成式奖励模型在强化学习中的性能瓶颈。该方法通过自竞争排序和锚定引导排序两种策略，将生成式奖励模型的比较性偏好转化为有效的标量奖励信号，从而提升大语言模型在开放式对话和推理任务上的强化学习效果。

Details

Motivation: 生成式奖励模型虽然在响应排序方面表现出色，但由于其比较性质与现有强化学习算法采用的标量评分范式不匹配，导致其在强化学习中未能充分发挥潜力。

Result: 在开放式聊天和推理基准测试上的实验表明，该方法显著提升了使用生成式奖励模型的强化学习训练效果，相比现有奖励构建方法取得了持续的性能增益。

Insight: 核心创新在于将生成式奖励模型的比较性输出（偏好排序）有效转化为强化学习可用的标量奖励信号，具体通过自竞争排序和锚定引导排序两种策略实现，为解决生成式奖励模型与强化学习范式不匹配的问题提供了新思路。

Abstract: Recent advances in reward modeling show a paradigm shift from discriminative reward models to generative reward models. However, despite their strong capabilities in response ranking, generative reward models have not realized their potential in reinforcement learning (RL). Our analysis reveals that this limitation arises from a mismatch between the comparative nature of generative reward modeling and the scalar scoring paradigm adopted by existing RL algorithms. To bridge this gap, we propose a Ranking-based Reward Construction (RRC) approach, which enables generative reward models to provide more effective RL learning signals by deriving rewards from relative preference rankings. RRC introduces two complementary strategies: self-competitive ranking, which exploits comparisons among sampled responses, and anchor-guided ranking, which enables scalable ranking-based reward construction with a small set of reference responses. Experiments across open-ended chat and reasoning benchmarks demonstrate that RRC substantially improves RL training with generative reward models, achieving consistent gains over existing reward construction approaches. Our code can be found at https://github.com/wangclnlp/RRC.

[95] CalibForge: Adversarial Solver Calibration for Scaling Learnable Terminal Tasks cs.LG | cs.CLPDF

Fanzhe Meng, Guoxin Chen, Jiale Zhao, Shuang Sun, Zhiyu Lin

TL;DR: 本文提出了CalibForge系统，这是一个用于自动合成终端任务的系统。它通过对抗性求解器校准，利用已验证的求解器行为来修订候选任务，旨在生成既可行又具有适当学习挑战性的任务。

Details

Motivation: 训练终端智能体需要可执行且可验证的任务，这些任务不仅要可解，还要对学习过程具有适当的挑战性。现有的可执行性验证仅能确立可行性，但无法揭示任务相对于特定求解器设置的行为特性。

Result: 使用CalibForge构建了5,431个校准后的终端任务。在Terminal-Bench 2.0基准测试上，训练出的模型分别达到32.58%和47.57%的性能。相比基线模型，在Terminal-Bench 2.0上最大提升24.71个百分点，在SWE-bench Pro上提升27.68个百分点，在Doc2Repo上提升30.04个百分点。

Insight: 核心创新在于提出了“求解器相对可学习性”作为构建有效训练数据的目标，并实现了两种对抗性校准策略（多求解器校准和对比求解器校准）来操作化这一概念。这为生成可迁移且有效的智能体训练数据提供了一种新方法。

Abstract: Training terminal agents requires executable and verifiable tasks that are not merely solvable, but appropriately challenging for learning. Executable validation establishes feasibility, yet does not reveal how a task behaves relative to a given solver setting. In this paper, we present CalibForge, an autonomous terminal-task synthesis system that uses verified solver behavior to revise candidate tasks through adversarial solver calibration. Multi-solver calibration targets disagreement within a heterogeneous solver pool, whereas contrastive solver calibration targets a designated strong-pass/weak-fail relation; both operationalize a solver-relative learnable zone anchored in demonstrated solvability. Using CalibForge, we construct 5,431 calibrated terminal tasks. Our ablations show that both strategies yield more effective supervision than authoring and validation alone or ordinary single-solver feedback. Models trained on the full collection achieve 32.58% and 47.57% on Terminal-Bench 2.0. The largest improvements over the corresponding base model reach 24.71 percentage points on Terminal-Bench 2.0, 27.68 points on SWE-bench Pro, and 30.04 points on Doc2Repo. Together, these results support solver-relative learnability as a practical target for constructing effective and transferable agent training data.

[96] Disentangling 3D Modeling from Spatial Reasoning cs.LG | cs.CVPDF

Haoze Sun, Jiequan Cui, Qingshan Xu, Richang Hong

TL;DR: 本文提出了一种新的空间推理范式DiSR，通过将3D感知与推理过程解耦，利用现成的感知模型重建结构化3D证据，并微调大语言模型仅基于这些几何证据进行推理。该方法在无需大规模3D视觉问答训练或复杂工具使用策略的情况下，在主流空间推理基准上取得了有竞争力的性能。

Details

Motivation: 动机在于观察到现代感知模型擅长估计连续3D几何，而大语言模型擅长组合与符号推理，因此希望结合两者优势，避免通过大规模训练隐式联合学习3D感知与推理。

Result: 在流行的空间推理基准测试中取得了有竞争力的性能，无需大规模3D VQA训练或复杂工具使用策略。

Insight: 创新点在于明确分离感知与推理的范式，利用专家模型提供显式几何证据供LLM推理，提高了可解释性、模块化和计算效率，为空间智能提供了一种可扩展的端到端建模替代方案。

Abstract: In this work, we explore an alternative paradigm for spatial reasoning by explicitly disentangling 3D perception from reasoning, rather than jointly acquiring implicit 3D perception and reasoning through large-scale training. Our key observation is that modern perception models excel at estimating continuous 3D geometry, whereas large language models (LLMs) are particularly effective at compositional and symbolic reasoning. Motivated by these complementary strengths, we propose the Disentangled Spatial Reasoner (DiSR), a simple yet effective framework that reconstructs the physical world into structured 3D evidence using off-the-shelf expert perception models and fine-tunes an LLM with LoRA to perform reasoning solely over this explicit geometric evidence. Without large-scale 3D VQA training or complex tool-use policies, DiSR achieves competitive performance on popular spatial reasoning benchmarks. Beyond its strong performance, DiSR offers improved interpretability, modularity, and computational efficiency, demonstrating that explicit separation of perception and reasoning is a scalable and effective alternative paradigm to end-to-end modeling for spatial intelligence.

[97] LC-GRPO: Bridging Train-Inference Gap for Flow-Based GRPO with Langevin Correction cs.LG | cs.AI | cs.CVPDF

Yingqing Guo, Hui Yuan, Zijian He, Mengdi Wang, Zheng Ding

TL;DR: 本文提出LC-GRPO，一种基于流的生成模型强化学习框架，通过引入朗之万校正来弥合训练时随机采样与推理时确定性ODE采样之间的差距。该方法在ODE欧拉步后添加一个朗之万校正步骤，以对准目标边际分布，无需额外分数模型。实验表明，该方法在文本到图像和文本到视频任务中能持续提升奖励优化效果，保持生成质量，并显著缩小训练与推理之间的差距。

Details

Motivation: 现有基于流的生成模型在强化学习训练中采用随机微分方程（SDE）进行探索，而推理时使用确定性常微分方程（ODE），导致两者离散化后的样本分布存在显著不匹配，SDE采样可能变得模糊，影响策略优化效果。

Result: 在SD3.5-Medium、FLUX.1-Dev和HunyuanVideo等基准测试中，LC-GRPO在文本到图像和文本到视频任务上一致地改进了奖励优化，保持了生成质量，并大幅缩小了随机训练与确定性推理之间的差距。

Insight: 创新点在于将朗之万校正步骤集成到基于流的GRPO框架中，通过ODE欧拉步与朗之万校正的组合，直接利用流速度恢复所需分数，无需额外模型，且过渡分布保持各向同性高斯，便于策略优化。理论分析表明，该方法能减少不完美ODE步的Wasserstein误差，并在相同随机性水平下比标准SDE离散化更准确。

Abstract: Flow-based generative models are typically sampled by solving a deterministic ordinary differential equation (ODE), whereas online reinforcement learning requires stochastic rollouts for policy exploration and optimization. Existing GRPO methods for flow models therefore replace the inference-time ODE with a stochastic differential equation (SDE) during training. Although the ODE and SDE share the same marginal distributions in continuous time, their finite-step discretizations can differ substantially. In particular, SDE rollouts often become blurry as the exploration noise increases, creating a mismatch between the samples used for reinforcement learning and those generated by the test-time ODE sampler. We introduce LC-GRPO, a flow-based GRPO framework with Langevin correction. Each rollout transition first takes an inference-aligned ODE Euler step and then applies a stochastic Langevin correction targeting the marginal distribution at the resulting timestep. The required score is recovered directly from the flow velocity, requiring no additional score model, while the resulting transition remains an isotropic Gaussian with a tractable likelihood for policy optimization. We theoretically show that, under suitable conditions, one Langevin correction step reduces the Wasserstein error of an imperfect ODE Euler step. At a matched randomness level, we further show that the proposed transition can be more accurate than the standard Euler–Maruyama discretization of the reverse SDE. Experiments on SD3.5-Medium, FLUX.1-Dev, and HunyuanVideo demonstrate that LC-GRPO consistently improves reward optimization across text-to-image and text-to-video tasks, preserves generation quality, and substantially narrows the gap between stochastic training rollouts and deterministic test-time ODE inference.

cs.GR [Back]

[98] GenGA: Editable and Data-Grounded Graphical Abstract Generation for Academic Papers cs.GR | cs.CL | cs.CV | cs.HC | cs.LG | cs.MMPDF

Takuro Kawada, Shunsuke Kitada, Hitoshi Iyatomi

TL;DR: 本文提出了GenGA框架，用于从学术论文内容生成可编辑的、数据驱动的图形摘要。该框架直接生成矢量格式的图形，支持导入现有绘图工具进行元素级编辑，并引入了结构独立性系数来量化图形的编辑简易性。

Details

Motivation: 现有方法生成的图形摘要多为栅格格式，难以进行文本修改和布局调整等后编辑，无法适应论文写作和同行评审中的迭代修订过程。

Result: 实验表明，GenGA在编辑简易性上优于传统方法，在简洁性和语义对齐方面甚至超越了人工绘制的图形摘要，同时验证了SIC指标与手动编辑成本的有效相关性。

Insight: 核心创新在于将图形摘要生成重新定义为基于研究者实际工作流的可编辑矢量图形生成问题，并提出了量化编辑简易性的SIC指标，显著促进了有效的科学交流。

Abstract: Graphical Abstracts (GAs) visually summarize the key findings of academic papers, playing a crucial role in facilitating the understanding of research content. Recently, advancements in vision-language models and image generation models have enabled the automatic generation of scientific figures based on paper content. However, most conventional methods output the generated results as raster graphics, making post-editing (e.g., text modification and layout changes) highly difficult. This poses a significant challenge, as they are unsuitable for the iterative figure revision process inherent in paper writing and peer review. To tackle these challenges, we define the novel task of generating editable GAs from paper content and propose GenGA, a new GA generation framework that directly produces figures in vector format. By generating figures as a collection of vector elements with a hierarchical structure, GenGA produces outputs that can be seamlessly imported into existing drawing tools for intuitive, element-level editing. Furthermore, we introduce the Structural Independence Coefficient (SIC), a metric that quantifies the editing simplicity of a figure based on the degree to which local modifications propagate to other elements. Experimental results show that GenGA achieves superior editing simplicity compared to conventional methods, and even surpasses human-authored GAs in conciseness and semantic alignment. We also validate SIC as an effective metric correlated with manual editing costs. This study fundamentally redefines GA generation as an editable vector graphic generation problem grounded in the practical workflows of researchers, significantly promoting effective scientific communication.

cs.CR [Back]

[99] A Survey of Adversarial Efficiency Degradation for Vision Transformer by Exploiting Input-adaptive Optimization cs.CR | cs.CVPDF

Anadi Goyal, Nandish Chattopadhyay, Anupam Chattopadhyay, Chandan Karfa

TL;DR: 这篇综述论文系统性地研究了针对视觉Transformer（ViT）输入自适应推理机制（如令牌剪枝和提前终止）的对抗性效率退化攻击。论文统一比较了两种代表性攻击方法（SlowFormer和DeSparsify）在三种流行令牌剪枝框架（A-ViT、ATS和AdaViT）上的表现，并提出了标准化的评估指标。

Details

Motivation: 随着ViT越来越多地依赖输入自适应推理来满足能效和延迟预算，针对这些效率优化机制的对抗性攻击成为一个新兴威胁。论文旨在系统理解这类攻击如何利用优化机制的攻击面来增加计算开销，而不一定降低模型精度。

Result: 论文通过标准化的GFLOPs、精度损失和攻击成功率（AS）指标对攻击进行了比较分析。攻击成功率衡量了攻击夺走了模型原本计算节省量的比例，为评估攻击效果提供了量化基准。

Insight: 论文的创新点在于首次系统性地将针对ViT效率优化机制的对抗性攻击进行统一分析和比较，并提出了标准化的评估框架。其核心洞察是揭示了输入自适应优化（如令牌剪枝）本身可能成为新的安全攻击面，这对在低功耗设备上部署高效模型时的防御设计具有重要启示。

Abstract: Vision Transformers (ViTs) increasingly rely on input-adaptive inference, such as token pruning and early halting, to meet energy and latency budgets. This survey examines a recent class of adversarial efficiency degradation attacks that target these mechanisms to increase computation without necessarily degrading accuracy. We unify and compare two representative attacks, SlowFormer (a universal adversarial patch) and DeSparsify (per-image perturbations), across three popular token-pruning frameworks: A-ViT, ATS, and AdaViT. We standardize reporting using GFLOPs, accuracy loss, and an Attack Success (AS) metric that measures how much of the model’s compute savings the attack takes away. Understanding these attacks is crucial for designing countermeasures that not only mitigate risk but also remain lightweight, since deployment often occurs in low-power settings such as mobile or embedded devices. To organize our analysis, we focus on three questions: how input-adaptive optimizations (e.g., token pruning and early halting) create attack surfaces for efficiency degradation; how such attacks operate in practice and which optimizations are most vulnerable; and which defenses exist today and whether they meaningfully restore efficiency under attack.

cs.AI [Back]

[100] Woodpecker Distillation: Weak Models Diagnose Reasoning Bugs in Strong Models cs.AI | cs.CLPDF

Dayu Wang, Jiaye Yang, Weikang Li, Jiahui Liang, Yang Li

TL;DR: 本文提出了一种名为Woodpecker Distillation的弱模型到强模型的训练框架，旨在修复大型语言模型在推理任务中出现的局部推理错误。该方法通过对比弱模型在相同前缀下生成的成功与不成功的补丁，构建一个纠正性的教师分布，并将其蒸馏到强模型中，从而提升其推理性能。

Details

Motivation: 论文的动机在于解决大型语言模型在具备解决能力的情况下，仍经常在推理任务中失败的问题。作者认为这些失败通常源于中间步骤的局部推理错误，而非全局能力不足，并且这些错误通常是可修复的。

Result: 在数学推理基准测试上的实验表明，Woodpecker Distillation能够持续提升强模型的性能，并且优于直接的模仿基线方法。

Insight: 论文的核心创新点在于提出了一种利用弱模型生成的对比性局部干预信号来指导强模型学习的方法，其关键洞察在于有效的学习信号并非来自干预文本本身，而是来自其对模型未来推理分布的修正作用。这为利用弱模型诊断和纠正强模型的推理错误提供了一种新范式。

Abstract: Large language models often fail on reasoning tasks despite possessing the capability to solve them. We argue that many such failures arise from localized reasoning bugs in intermediate steps rather than from global incompetence. We show that these bugs are frequently repairable: inserting a short patch generated by a weak probe model after the same strong-model reasoning prefix can redirect the trajectory toward a correct solution. However, this corrective effect is not reliably internalized by directly fine-tuning on weak patches or repaired trajectories, suggesting that the useful signal lies not in the intervention text itself, but in how it reshapes the model’s future reasoning distribution. We therefore propose Woodpecker Distillation, a weak-to-strong training framework that learns from contrastive local interventions. Our method contrasts successful and unsuccessful weak-model patches at the same prefix, constructs a corrective teacher distribution from their induced future token predictions, and distills this signal into the strong model. Experiments on mathematical reasoning benchmarks show that Woodpecker Distillation consistently improves strong-model performance and outperforms direct imitation baselines.

[101] Refining Over Resampling: Test-Time Self-Correction for LLM Reasoning cs.AI | cs.CLPDF

Ahsan Bilal, Muhammad Ahmed Mohsin, Muhammad Umer, Lena Trigg, Ali Subhan

TL;DR: 本文提出了一种无需验证器的广度-深度精炼框架，用于提升大语言模型在推理任务中的测试时性能。该方法通过采样多个独立推理路径、对每条路径进行迭代式自我批判与修正，并采用多数投票聚合答案，从而在探索多样性的同时修复局部推理错误。

Details

Motivation: 现有测试时扩展方法（如增加采样数量或依赖验证器选择）存在收益递减或校准依赖问题，无法有效利用额外计算资源来提升推理质量。

Result: 在AIME24、AIME25、AMC、OlympiadBench和MATH500等多个数学推理基准上，该方法均优于贪婪解码、多数投票、基于验证器的最佳选择、束搜索和前向解码等基线。例如，使用Qwen2.5-1.5B模型时，在MATH500上的准确率从最强验证器基线的58.0%提升至32.5%（AMC从25.0%提升至32.5%）。

Insight: 核心创新在于将测试时计算资源用于对采样轨迹进行迭代精炼（自我批判与修正），而非仅用于增加采样数量或依赖外部验证器。这通过结合广度（保持初始尝试多样性）和深度（修复局部错误）的策略，更有效地利用了推理计算。

Abstract: Test-time scaling improves LLM reasoning by using additional inference compute, but wider sampling alone can suffer from diminishing returns: new rollouts often repeat existing answer patterns instead of adding useful reasoning diversity. Verifier-based selection offers an alternative, but its performance depends on the calibration of an external reward model. We propose a verifier-free breadth–depth refinement framework that uses test-time compute to both explore and improve candidate solutions. The method samples multiple independent reasoning rollouts, refines each rollout through iterative self-critique and self-correction, and aggregates the refined answers by majority voting. Breadth preserves diverse initial attempts, while depth repairs local reasoning errors before aggregation. Across AIME24, AIME25, AMC, OlympiadBench, and MATH500, our method consistently improves over greedy decoding, majority voting, verifier-based best-of-$N$, beam search, and lookahead decoding across multiple open-weight models. For instance, with Qwen2.5-1.5B, accuracy increases from the strongest verifier-based baseline to $58.0%$ on MATH500, and from $25.0%$ to $32.5%$ on AMC. These results show that test-time compute can be more effective when used to refine sampled trajectories rather than only to sample more candidates or rely on verifier-guided selection.

[102] Unified Agent: Managing Interactions across Devices cs.AI | cs.CL | cs.CV | cs.HCPDF

Xinshuang Liu, Runfa Blark Li, Shaoxiu Wei, Xin Lin, Truong Nguyen

TL;DR: 本文提出了一种名为Unified Agent的状态管理智能体，旨在解决跨设备、跨时间交互场景中现有智能体系统状态管理不足的问题。该智能体通过维护一个紧凑、可行动的状态来整合历史交互证据、陈述事实和待处理请求，从而基于当前观察做出决策。

Details

Motivation: 现有智能体系统在跨设备、跨时间交互场景中表现不佳，因为观察信息分散在不同设备和时刻，而主流系统（如单一工具型智能体或多智能体系统）缺乏有效的状态管理机制来整合这些信息。

Result: 在构建的跨设备、跨时间用户-智能体交互基准测试中，Unified Agent在默认设置下显著优于四种已发表设计的适配版本。即使更换多模态大语言模型（MLLM）家族、能力或推理资源，其性能始终领先于所有对比系统，表明状态设计优势在不同MLLM设置下具有鲁棒性。

Insight: 创新点在于提出了以状态为中心的设计原则，将分散的交互证据、事实和请求组织成紧凑、可行动的状态，从而提升跨设备、跨时间交互的决策效率。客观来看，这种状态管理机制为智能体在复杂、分散环境中的长期交互提供了可借鉴的架构思路。

Abstract: As capabilities rapidly increase, AI agents can move from running inside one app to acting across a user’s devices over time. Yet existing agent systems still fall short in this scenario. This is because observations are scattered across devices and moments, but mainstream systems are not designed around this fact: a single agent that treats devices as tools lacks effective state management for all devices across time, and multi-agent systems coordinate across agents but do not maintain the compact carried state a cross-device, cross-time request needs. We argue that the agent should maintain an effectively designed state that organizes engagement evidence, stated facts, and the standing request in a compact, action-ready form for deciding its action given the current observation. To compare state designs, we construct a benchmark of user-agent interaction across devices and time. We instantiate this principle in Unified Agent, a stateful agent that carries interaction evidence across devices and moments and uses it with the current observation to act. In the default setting, it significantly outperforms our adaptations of four published designs. Across changes in multimodal large language model (MLLM) family, capability, and reasoning effort, it remains ahead of all compared systems, demonstrating that the state-design advantage is robust across MLLM settings. Our code and data will be publicly available on GitHub.

[103] When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in LLM Agents cs.AI | cs.CLPDF

Linfang Shang, Ming Xu, Yiding Sun, Tianle Xia, Lingxiang Hu

TL;DR: 本文研究了大型语言模型（LLM）智能体在自我进化过程中，通过从执行轨迹中提炼可重用技能来积累能力时，可能出现的性能退化现象。研究发现，当技能池超过临界规模后，新添加的技能反而会污染决策上下文，导致性能下降，且这种污染在结构上是不可逆的。为此，论文提出了“验证者即守门员”（VaG）方法，通过三层异构验证器在技能入库前进行过滤，并结合边际增益子集选择来防止组合污染，从而实现了持续的性能提升和技能池的正向迁移。

Details

Motivation: 动机在于发现LLM智能体自我进化过程中的非单调性：技能的无条件积累在超过临界点后会导致性能下降，即“技能污染”。这源于有缺陷的技能一旦进入决策上下文，会成为后续技能提炼的参考材料，形成跨轮次的污染链，且事后移除污染源无法完全恢复性能。

Result: 在Terminal-Bench 2基准测试上，无条件技能积累的性能先升后降，事后移除污染技能仅能恢复一小部分性能损失，验证了污染的不可逆性。相比之下，VaG方法每轮性能均提升，以约小5倍的技能池规模达到了72%的pass@1成功率，并且其冻结的技能池能正向迁移到其他四个骨干模型和第二个基准测试上，无需重新进化。消融实验证实了三个验证器的互补性和不可替代性。

Insight: 核心创新点在于将技能准入视为必须预先承诺（pre-commit）而非事后修复的问题，并提出了VaG框架。该框架通过结构有效性、行为无害性和语义一致性三个异构验证器在技能入库前进行分层过滤，并结合边际增益子集选择来主动防止组合污染，从而在结构上保证了技能池的质量和进化过程的鲁棒性。这为构建可扩展且可靠的自我进化智能体提供了新的设计原则。

Abstract: Self-evolving agents accumulate capability by distilling reusable skills from their execution trajectories, but we find this process is not monotonic: past a critical pool size, newly added skills degrade performance instead of improving it. We formalize this capability-contamination phase transition and trace it to a structural cause: once a defective skill enters the decision context, it becomes reference material for distilling later skills, forming cross-round contamination chains. We further show the contamination is structurally irreversible: removing a source skill after the fact cannot erase the flawed reasoning its descendants have already inherited, so post-hoc rollback recovers only a small fraction of the lost performance. This makes skill admission a pre-commit necessity rather than a post-hoc fix, and motivates Verifier-as-Gatekeeper (VaG): a progressive trust hierarchy whose three heterogeneous critics - structural validity, behavioral harmlessness, and semantic consistency - filter each skill individually, coupled with a marginal-gain subset selection that removes combinatorial contamination at the top tier before skills reach the runtime context. On Terminal-Bench 2, unconditional accumulation rises to a peak and then degrades, giving back most of its gains as the pool keeps growing, and post-hoc removal of the culprit skills recovers only a small part of the drop - the empirical signature of irreversibility. In contrast, VaG improves every round, reaching 72% pass@1 with a pool roughly 5x smaller, and its frozen skill pool transfers positively to four other backbones and a second benchmark without re-evolution. Ablations confirm the three critics are complementary and mutually non-substitutable, each intercepting a largely disjoint class of harmful skills.

[104] AppDeltaWorld: Transition-Grounded Delta Code World Model for Mobile GUI Agents cs.AI | cs.CLPDF

Weikai Xu, Yunren Feng, Haoxiang Lei, Kun Huang, Yuxuan Liu

TL;DR: 本文提出了AppDeltaWorld，一种基于转换约束的增量代码世界模型，用于移动GUI代理。该模型将下一个GUI预测为可达的代码更新，而非无约束的图像或文本描述，通过检索应用特定的HTML参考、生成可执行的HTML并插入视觉资源来实现。作为世界模型，它在CMGUIBench-500上实现了最高保真度；作为训练环境，它支持过滤闭环SFT数据构建，使AppDeltaAgent在AndroidLens上达到SOTA性能，并在MobileGym和MobileWorld上获得一致提升。

Details

Motivation: 解决移动GUI代理在真实轨迹难以获取、现有模拟环境扩展成本高以及GUI世界模型存在生成不稳定、模态覆盖有限和动作-转换逻辑不一致等问题。

Result: 在CMGUIBench-500的Code2World评估中达到最高保真度，在结构布局和UI元素重建上优于仅图像和仅代码基线；作为训练环境，AppDeltaAgent在AndroidLens上实现SOTA性能，在MobileGym和MobileWorld上获得一致增益，且基于世界模型的测试时强化学习能进一步改进策略而无需与真实应用交互。

Insight: 创新点在于将GUI预测建模为可达的增量代码更新，结合检索、生成和渲染流程，提高了生成稳定性和模态覆盖；客观分析认为，该方法通过约束动作-转换逻辑，有效解决了现有世界模型的不一致问题，为移动交互策略训练提供了可扩展的模拟环境。

Abstract: Mobile GUI agents can operate apps through pixel perception and touch actions, making them a promising interface for collecting and improving long-horizon mobile interaction policies. However, real trajectories are difficult to obtain for sensitive apps and privacy-critical operations. At the same time, existing simulated environments are costly to scale up, and GUI world models still suffer from unstable generation, limited modality coverage, and inconsistent action-transition logic. To address these limitations, we propose AppDeltaWorld, a transition-grounded delta code world model that predicts the next GUI as a reachable code update rather than as an unconstrained image or text description. AppDeltaWorld retrieves app-specific Level-1 HTML references under an action-transition constraint, generates Level-2 executable HTML conditioned on the current screen, action, predicted next-screen text, and retrieved structure, and inserts generated visual assets into image slots before browser rendering. As a world model, AppDeltaWorld achieves the highest fidelity on CMGUIBench-500 under Code2World evaluation, with clear gains in structural layout and UI element reconstruction over image-only and code-only baselines. As a training environment, AppDeltaWorld supports filtered closed-loop SFT data construction that, when combined with public supervision, enables AppDeltaAgent to achieve state-of-the-art performance on AndroidLens and consistent gains on MobileGym and MobileWorld. Moreover, world-model-based test-time reinforcement learning enables policy adaptation and shows further improvements without additional interaction with real apps.

[105] Poli-Bias: Understanding and Measuring Large Language Model Biases in International Political Conflicts cs.AI | cs.CLPDF

Massi-Nissa Abboud, Aladin Djuhera, Elena Cabrio, Holger Boche

TL;DR: 本文提出了Poli-Bias框架，用于测量大型语言模型在国际政治冲突中的偏见。该框架通过系统性地交换提示中的国家身份，在成对的、法律上等效的冲突场景中比较模型的响应，从而揭示模型是否存在因国家身份或用户立场不同而产生的系统性差异。

Details

Motivation: 动机在于，LLMs中的政治偏见难以衡量，因为它可能通过微妙的措辞、论证和法律推理差异来体现，而单一的指标难以捕捉这些复杂表现。

Result: 在涵盖不同模型系列和规模的13个当代LLMs上的评估表明，国家身份和用户立场会系统性地影响模型对等效行动在国际法下的描述、评估和辩护。

Insight: 创新点在于提出了一个反事实的、细粒度的审计框架，将响应差异分解为五个可解释的维度，以揭示不平等对待的具体表现方式和位置，而不是将偏见简化为单一判断。

Abstract: Measuring political bias in large language models (LLMs) remains challenging as it can manifest through subtle differences in framing, argumentation, and legal reasoning that are difficult to capture with a single metric. In this work, we introduce Poli-Bias, a counterfactual framework for measuring whether LLMs treat legally equivalent conflict scenarios differently depending on the countries involved. Poli-Bias compares responses to paired prompts in which country identities are systematically swapped across diverse geopolitical relationships, legal violations, and reasoning tasks. Rather than reducing bias to a single judgment, our framework decomposes response disparities into five interpretable dimensions, revealing how and where unequal treatment manifests. Across 13 contemporary LLMs spanning diverse model families and sizes, we find that country identities and user affiliations can systematically affect how equivalent actions are described, evaluated, and defended under international law. Our results thus establish Poli-Bias as a fine-grained framework for auditing political even-handedness and sycophancy in LLMs.

[106] ECHO: A Locally-Deployable Agentic Health Assistant with Temporal Memory, Safety Guardrails, and Speech Assessment cs.AI | cs.CLPDF

Abdulkadir Külçe, Alihan Esen, Cağla Fikir, Berke Kurt, Kuzey Arar

TL;DR: 本文提出了ECHO（增强护理与健康观察者），一个可本地部署的对话式健康助手，用于长期慢性病管理。该系统集成了三个互补的软件模块：一个基于ReAct循环和LangGraph构建的智能聊天机器人，配备临床工具和时序知识图谱以实现跨会话记忆；一个两阶段混合安全层，用于拦截和分类用户查询；以及一个结合语音和文本的多模态语音评估模块。整个系统以Web应用形式实现，可在消费级硬件上本地运行，无需将患者数据传输至外部服务。

Details

Motivation: 解决长期慢性病管理中，需要一个能够提供持续、安全、私密且具备记忆能力的对话式健康助手的需求，同时满足数据隐私法规（如GDPR、KVKK）的要求。

Result: 核心聊天机器人在59个场景的基准测试中，工具执行通过率达到94.9%（使用GPT-5 Mini）。安全层在2537条标注的土耳其健康查询数据集上，对边界案例的临床意图分类准确率达到88.8%，不安全查询召回率达到90.6%，优于包括Llama 3.3 70B在内的零样本LLM基线。语音评估模块在情感、抑郁和疼痛估计上的平均宏观F1分数为0.652。

Insight: 创新点在于将智能体架构（ReAct/LangGraph）、时序知识图谱、混合安全机制（规则+图神经网络）和多模态语音评估集成到一个可本地部署的统一系统中。其模块化设计、对数据隐私的本地化处理以及针对临床安全性的专门优化，为构建安全、合规的医疗对话系统提供了可借鉴的框架。

Abstract: This paper presents ECHO (Enhanced Care & Health Observer), a locally-deployable conversational health assistant for long-term chronic care management. ECHO integrates three complementary software modules developed under shared supervision as a unified system. The core module is an agentic chatbot built on a ReAct loop orchestrated via LangGraph, equipped with 17 clinical tools and a temporal knowledge graph for persistent cross-session memory; it achieves a 94.9% tool-execution pass rate across a 59-scenario benchmark with GPT-5 Mini. A two-stage hybrid safety layer intercepts all incoming queries: a rule-based layer handles explicit crisis signals and jailbreak attempts in under 1ms, while a signed graph neural network (GNN) with APPNP-style propagation classifies boundary cases by clinical intent, achieving 88.8% accuracy and 90.6% unsafe recall on a 2,537-query annotated Turkish health dataset while outperforming zero-shot LLM baselines including Llama 3.3 70B. A multimodal speech assessment module combining Whisper acoustic encoding and BERT text encoding with cross-attention fusion estimates emotion, depression, and pain, reaching a mean macro F1 of 0.652. The full system is implemented as a web application that can run entirely on consumer hardware, with no patient data transmitted to external services, supporting compliance with GDPR and KVKK.

[107] Coherence-Oriented Dream Scene Visualisation cs.AI | cs.CVPDF

Azra Açıl, Simon Colton

TL;DR: 本文介绍了Dream Scene Visualiser (DSV)系统，该系统能够将书面的梦境描述转化为由四幅面板图像组成的时间序列，以可视化梦境。系统首先利用大语言模型将梦境描述按时间顺序分割为四个部分，然后使用文本到图像模型为每个部分生成图像，并确保序列间的视觉连贯性，同时DSV会重新生成任何与文本不匹配的图像。

Details

Motivation: 解决梦境情感强烈但难以沟通的问题，旨在通过自动化系统将文字梦境描述转化为连贯的视觉序列。

Result: 在DreamBank的50个梦境描述上进行评估，使用CLIP、DINOv2和Qwen2-VL等视觉-语言模型进行客观测量，报告了质量、保真度和连贯性结果。

Insight: 创新点在于结合大语言模型进行时序分割和文本到图像模型生成，并通过再生机制确保图像与文本的匹配，实现了梦境描述的连贯视觉化；从客观角度看，该方法在视觉连贯性维护和多模态评估方面具有借鉴意义。

Abstract: Dreams can be emotionally intense but difficult to communicate. We describe the Dream Scene Visualiser (DSV) system which turns written dream descriptions into a temporal sequence of four panel images visualising the dream. This starts with a large language model prompted to split a dream description into four chronological parts. Then a text-to-image model produces images for each part with visual coherence maintained across the sequence, and DSV regenerates any image not suitably matching the text. We evaluate DSV over 50 visualisations from dream descriptions in DreamBank, and report quality, fidelity and coherence results via objective measures employing the CLIP, DINOv2 and Qwen2-VL vision-language models.

[108] Training a Conditioned Video Game Agent on a VLM Annotated Dataset cs.AI | cs.CV | cs.LGPDF

Katrin Schmid, Iuri Frosio

TL;DR: 本文提出一种利用视觉语言模型（VLM）标注视频游戏数据集以提取人工定义奖励的方法，旨在简化强化学习在游戏智能体训练中的应用。通过离线强化学习，训练出一个能根据期望回报做出响应的条件化智能体，并讨论了早期实验中的困难与局限。

Details

Motivation: 解决传统强化学习在视频游戏策略学习中面临的三大难题：需要访问游戏引擎获取奖励、奖励函数的识别与加权需反复试错、奖励稀疏且对策略的影响难以理解。

Result: 论文表明离线强化学习可用于训练条件化智能体，使其能根据VLM标注的奖励做出响应，但未提及具体基准测试或定量结果，仅讨论了早期实验的困难与局限。

Insight: 创新点在于利用VLM自动标注游戏数据集以生成人类可理解的奖励信号，从而绕过传统RL对游戏引擎的依赖和奖励设计的复杂性，为可解释且易用的策略学习提供了新思路。

Abstract: Reinforcement Learning (RL) is a powerful but far from easy-to-use technique for policy learning. In the specific case of video games, access to the game engine is required to get rewards for training (e.g. to collect rewards from the environment). Furthermore, the proper identification and weighting of the rewards generally requires a difficult trial-and-error approach. Lastly, rewards are often sparse and understanding how they eventually affect the learned policy is a non-trivial exercise. To ease these issues we propose annotating a video game dataset with Vision Language Models (VLMs) instructed to extract human defined rewards. We show that offline RL can then be used to train a conditioned agent that responds accordingly to the desired returns and we discuss the difficulties and limitations that emerged in our early experiments.

[109] GAUGE: A Measurement-Grounded Benchmark for Physical Fidelity in Simulation Engines and Video World Models cs.AI | cs.CV | cs.ROPDF

Shuai Wang, Yaxin Feng, Xuekun Jiang, Shihan Tian, Ningyu Yan

TL;DR: 本文提出了GAUGE基准，这是一个基于真实世界测量的诊断性基准，用于联合评估数值模拟器（物理引擎）和生成式视频世界模型在再现真实世界物理规律方面的保真度。该基准包含22个受控任务族，涵盖刚体、柔性线缆、纺织品和体积可变形物体，并基于真实轨迹、校准的物理元数据、不确定性标注和任务特定观测值。研究对Isaac Sim、Genesis和Newton物理引擎进行了14个任务族的评估，并对6个图像到视频模型在5个刚体任务上进行了测试，重点关注物理定律一致性和推断参数的时间稳定性。

Details

Motivation: 现有对物理引擎和视频世界模型物理保真度的评估通常是孤立的，且严重依赖感知相似性或人类判断，难以揭示具体违反了哪些物理原理或参数。因此，需要一种基于真实世界测量、能够诊断物理偏差的联合评估基准。

Result: 评估结果显示，没有物理引擎在所有方面都完全忠实于物理，最大的偏差出现在脉冲接触、快速纺织品运动和体积变形中。视频世界模型虽然能产生符合预期方程形式的轨迹，但在恢复正确的加速度、动量传递和振荡时序方面存在错误。

Insight: 论文的创新点在于提出了一个测量驱动的、多模态（模拟器与生成模型）的物理保真度联合诊断基准，其任务设计根植于真实世界数据并覆盖广泛的物理过程和材料，为开发更符合物理的模拟器和世界模型奠定了基础。从客观角度看，其将物理引擎与生成模型置于同一框架下评估的思路，以及对物理原理进行细粒度诊断的方法，具有借鉴意义。

Abstract: Physics engines facilitate large-scale training and evaluation for embodied intelligence, while generative video world models are emerging as implicit simulators of future states and interactions. However, existing evaluations of physical fidelity are often conducted in isolation and rely heavily on perceptual similarity or human judgments, providing limited insight into which physical principles or parameters are violated. We introduce GAUGE, a real-world-grounded diagnostic benchmark for jointly evaluating how numerical simulators and generative video world models reproduce or deviate from real-world physics. It comprises 22 controlled task families covering rigid bodies, flexible cables, textiles, and volumetric deformable objects. Grounded in real-world trajectories and paired with calibrated physical metadata, uncertainty annotations, and task-specific observables, these tasks cover fundamental physical processes including collision, friction, momentum transfer, oscillation, self-contact, and deformation across diverse materials and conditions. We benchmark Isaac Sim, Genesis, and Newton on 14 task families using generalized trajectory errors, and evaluate 6 image-to-video models on 5 rigid-body tasks by testing physical-law consistency and the temporal stability of inferred parameters. Our results reveal no uniformly faithful physics engine, with the largest discrepancies arising in impulsive contact, rapid textile motion, and volumetric deformation. We further find that video world models can produce trajectories with the expected equation form while recovering incorrect accelerations, momentum transfer, and oscillation timing. GAUGE lays the groundwork for developing more physically faithful simulators and world models for embodied intelligence.

[110] VLMs for Videogame Data Annotation cs.AI | cs.CV | cs.LGPDF

Katrin Schmid, Iuri Frosio

TL;DR: 本文研究了视觉语言模型（VLMs）在视频游戏帧序列标注奖励信号任务中的应用，发现VLMs在赛车类游戏等合成场景中表现不佳，并探讨了输出混合、提示优化等应对策略，以及输入序列长度、分辨率和问题批处理对标注质量和计算消耗的影响。

Details

Motivation: 解决VLMs在视频游戏这种合成场景多变且不符合真实世界物理规律的环境中应用受限的问题，探索其用于游戏帧序列奖励标注以支持条件训练和离线强化学习等任务。

Result: 论文表明VLMs在赛车游戏等类型中难以回答基本问题，通过实验分析了输入参数对标注质量的影响，但未提及具体基准测试或与现有方法的定量比较结果。

Insight: 创新点在于将VLMs应用于视频游戏数据标注这一新领域，并针对其局限性提出了输出混合和提示优化等实用对策，为合成场景下的VLM适应性研究提供了见解。

Abstract: Vision Language Models (VLMs) and Artificial Intelligence (AI) agents have revolutionized how engineers approach complex problems in real-world applications. Their adoption in video games is on the other hand limited by the extreme variability of the synthetic scenarios and their poor compliance with real-world physics. Here we investigate the use of VLMs for annotating video game frame sequences with reward signals, a task with several potential applications including, among others, conditioned training and offline reinforcement learning. We show that VLMs often struggle to answer basic questions on racing video games (although we observed a similar behavior on other game genres) and discuss countermeasures such as VLM output mixing and prompt optimization. We also show how input sequence length, resolution, and question batching affect the annotation quality and its token consumption.

cs.SE [Back]

[111] LangChoiceBench: Measuring and Explaining Programming-Language Choice in LLMs cs.SE | cs.CLPDF

Lukas Twist, Twm Stone, Helen Yannakoudakis, Jie M. Zhang

TL;DR: 本文介绍了LangChoiceBench，一个用于评估大型语言模型在项目级代码生成中编程语言选择偏好的基准测试。该基准旨在系统测量模型的Python偏好、推荐与实现一致性以及语言多样性。研究发现，现有LLMs普遍过度选择Python，且推荐与实现一致性较低，同时揭示了模型选择Python时存在自动化倾向、虚构证据等失败模式。

Details

Motivation: 针对当前缺乏系统方法评估LLMs在项目级代码生成中编程语言选择行为的问题，特别是Python偏好现象，研究者旨在开发一个标准化基准来量化这一行为。

Result: 在覆盖7个软件领域28个项目的LangChoiceBench上评估了25个LLMs，结果显示Python被严重过度选择，推荐与实现一致性低，且较小开源模型表现出更强的Python偏好和更低语言多样性。通过分析9,826条推理轨迹，发现了自动化选择、虚构证据等具体行为模式。

Insight: 创新点在于构建了首个系统测量LLMs编程语言选择行为的基准，并深入分析了其决策机制，揭示了‘幻影证据’这一重要的模型失败模式，为理解模型在代码生成中的偏见提供了新视角。

Abstract: Large language models (LLMs) have been shown to exhibit strong Python preferences when generating project-level code, but there is currently no systematic way to measure this behaviour across new models. To bridge this gap, we introduce LangChoiceBench, a project-level code-generation benchmark for measuring Python preference, recommendation-implementation consistency, and language diversity. LangChoiceBench covers 28 projects across seven software areas where Python is often a poor default. We evaluate 25 diverse LLMs and find that Python remains heavily over-selected, recommendation-implementation consistency is low, and smaller open-weight models generally show stronger Python preference and lower language diversity. We further analyse 9,826 reasoning traces and find that most Python choices are automatic or driven primarily by ease, rather than explicit consideration of project requirements. In a smaller but important set of cases, models fabricate contextual support for choosing Python - a failure mode we call phantom evidence - or produce code that contradicts the language selected in their own reasoning.

cs.RO [Back]

[112] VLAff: Vision-Language-Affordance Model for Unified Actionable Affordances cs.RO | cs.CVPDF

Jihoon Oh, Kento Kawaharazuka, Kei Okada

TL;DR: 本文提出VLAff模型，一个基于视觉-语言大模型的统一基础模型，用于学习可操作的功能可供性。该模型利用人类第一人称视角视频，结合先进的3D运动结构和手部网格重建技术，提取视觉、抓取和轨迹三种可供性。研究构建了大规模数据集EgoAffordance，并通过实验验证了VLAff在视觉可供性预测上达到SOTA，并能有效应用于零样本机器人操作等任务。

Details

Motivation: 从人类视频中学习操作技能对可扩展的机器人学习具有前景，但人体与机器人之间的形态差异带来了挑战。本文旨在学习与形态无关的、以物体为中心的可操作功能可供性来解决这一问题。

Result: VLAff在视觉可供性预测任务上取得了最先进的性能。实验表明，该模型能有效应用于真实机器人任务，如零样本操作和可供性引导的机器人学习。

Insight: 创新点在于提出了一个统一的、基于视觉-语言大模型的基础框架，能够联合学习多种可操作可供性（视觉、抓取、轨迹）的跨模态关联，并构建了大规模数据集EgoAffordance来支持模型训练。从客观角度看，将多种可供性预测统一到一个模型中，并结合3D场景信息生成可直接执行的动作，是一个有前景的系统性方法。

Abstract: Learning manipulation skills from human videos is promising for scalable robot learning. However, the embodiment mismatch between humans and robots makes this challenging. One promising solution is to learn object-centric actionable affordances that are embodiment-agnostic. In this work, we propose a framework that leverages egocentric human videos with state-of-the-art 3D Structure-from-Motion and hand mesh reconstruction to extract actionable affordances such as visual, grasp, and trajectory affordances that explicitly encode where to interact, how to grasp, and how to move. We construct EgoAffordance, a large-scale dataset comprising 204K episodes with 5.6M visual affordances and 11.6M grasp and trajectory affordances. Building on this, we introduce VLAff, a large vision-language model-based unified foundation model that learns cross-modal correlations across all actionable affordances. Given a visual observation and instruction, VLAff generates visual affordance heatmaps, grasp poses, and trajectories, which are then converted into directly executable actions by utilizing 3D scene information. Through extensive experiments, we demonstrate that VLAff not only achieves state-of-the-art performance on visual affordance prediction, but can also be effectively applied to real robot applications such as zero-shot manipulation and affordance-guided robot learning.

[113] World-to-Wrist: Task-Conditioned Future Wrist Modeling for Fine-Grained Robot Manipulation cs.RO | cs.CVPDF

Yuhao Pan, Haosong Peng, Zhengshen Zhang, Zhengyang Yan, Yalun Dai

TL;DR: 本文提出World-to-Wrist VLA (W2-VLA)模型，一种用于精细机器人操作的视觉-语言-动作模型，其核心创新在于引入了任务条件化的未来手腕建模。模型通过一个潜在建模令牌接口连接视觉语言模型与手腕预测器，在给定多视角观测和任务指令下，预测未来手腕潜在状态以生成未来感知的动作上下文。此外，作者还提出了W2-CoT合成流程来生成结构化标注，为潜在接口提供辅助监督。

Details

Motivation: 现有视觉-语言-动作模型通常将主视角和手腕视角观测视为平行的视觉输入，忽略了它们在机器人操作中的不同作用。精细操作需要根据全局任务上下文来预测手腕局部交互的可能演变，现有方法存在这一局限。

Result: 在LIBERO、RoboTwin 2.0和真实世界操作任务上的实验表明，该方法在单臂和双臂设置下均提升了精细化和接触敏感的操作性能，同时保持了高于80 Hz的动作生成速率。

Insight: 主要创新点在于将任务条件化的未来手腕建模显式地整合到VLA框架中，通过一个紧凑的潜在接口桥接全局任务理解与局部手腕动态预测。W2-CoT提供的结构化标注（描述操作进度、物理转换线索和手腕局部证据）作为一种新颖的辅助监督形式，有助于塑造这一接口，是可借鉴的思路。

Abstract: Vision-language-action (VLA) models often treat main-view and wrist-view observations as parallel visual inputs, overlooking their distinct roles in robot manipulation. Fine-grained manipulation, however, benefits from anticipating how wrist-local interactions may evolve under the global task context. To address this limitation, we present World-to-Wrist VLA (W2-VLA), a VLA model for fine-grained robot manipulation with task-conditioned future wrist modeling. Given current multi-view observations and a task instruction, W2-VLA contextualizes a set of latent modeling tokens as a compact interface between the vision-language model and the wrist predictor. Conditioned on this interface and the observed wrist history, the predictor forecasts future wrist latents, which are transformed into future-aware context for action prediction. In addition, we introduce W2-CoT, a synthesis pipeline that produces structured annotations describing manipulation progress, physical transition cues, and wrist-local evidence. These annotations provide auxiliary supervision that shapes the task-conditioned latent interface. Experiments on LIBERO, RoboTwin 2.0, and real-world manipulation tasks demonstrate improved fine-grained and contact-sensitive manipulation across both single-arm and bimanual settings, while maintaining action-generation rates above 80 Hz.

[114] Visual Grounding in Zero-Shot Vision-Language Control cs.RO | cs.AI | cs.CVPDF

J. de Curtò, Dayani Plasencia, Diego Sánchez, I. de Zarzà

TL;DR: 该论文研究了视觉语言模型（VLMs）在零样本视觉语言控制中的视觉基础能力，通过输入消融测试发现现有直接控制模型在视觉基础方面存在缺陷，但提出了一种后处理对称一致性守护机制，能够筛选出可靠的模型并实现高精度的危险检测和镜像等变性。

Details

Motivation: 动机在于验证VLMs作为零样本控制器时是否真正基于视觉输入进行决策，而非依赖模拟器动力学或保守动作先验，从而揭示其视觉基础能力的局限性。

Result: 在多个模拟器和模型上的大规模测试表明，直接控制模型表现不佳，但提出的守护机制在保留帧上达到0.954平衡准确率（95% CI [0.895,0.990]），离线模块重放实现0.934动作一致性和精确镜像等变性。

Insight: 创新点在于通过系统性的输入消融测试揭示了VLMs视觉基础的模块化失败，并设计了一种后处理对称一致性守护机制，能够有效筛选可靠模型，支持将VLMs视为有界的选择性危险助手而非整体零样本控制器。

Abstract: Vision-language models (VLMs) are increasingly used as zero-shot controllers, but successful trajectories do not necessarily show that decisions are grounded in visual input: simulator dynamics and conservative action priors can produce favourable scores without meaningful perception. We investigate this with an input-ablation battery: blind-image controls, repeated identical inputs, lane-axis reflection, non-visual baselines, and pipeline-integrity checks. Across nine direct-action models, six structured local VLMs, and an exploratory VLM-MPC hierarchy, we analyse 32,874 scored calls over two embodiments and three simulators. The direct-control results are largely negative: a constant-SLOW policy outperforms a scripted geometric controller, several models are image-invariant or nearly constant, and models that recognize longitudinal hazards still fail to transform LEFT and RIGHT under reflection. No local VLM meets the joint longitudinal and lateral grounding criteria. However, an image-only deterministic positive control estimates the lead gap with 0.090 m MAE and exact mirror equivariance, confirming the stimuli carry sufficient visual information; the failures are modular, not universal. A post-hoc, leakage-controlled symmetry-consensus guardian selects two models from 16 calibration frames and freezes a 2-of-4 hazard vote across original and reflected views. On 272 held-out frames it reaches 0.954 balanced accuracy (episode-cluster bootstrap 95% CI [0.895,0.990]); nested leave-one-episode-out recovers the same pair and threshold in all 12 folds. Abstaining on ties raises committed balanced accuracy to 0.973 at 0.824 coverage. With deterministic perception retaining lateral authority, offline modular replay achieves 0.934 action agreement and exact mirror equivariance. These results support current VLMs as bounded, selective hazard assistants, not monolithic zero-shot controllers.

[115] Prior-SG: Task and Prior Driven Region Segmentation for Scene Graphs in Arbitrarily-Structured Environments cs.RO | cs.CVPDF

Giorgio Tonetti, Laurent Kneip, Abel Gawel, Marco Hutter

TL;DR: 本文提出Prior-SG框架，用于在任意结构环境中生成3D场景图。该框架将场景图生成视为概率对齐问题，通过融合多尺度开放词汇特征构建实例图，并利用大语言模型动态合成的先验图指导最大后验估计，以推断高层功能语义。

Details

Motivation: 现有3D场景图提取框架依赖局部视觉聚类或严格几何启发式方法（如以墙分隔房间），在开放式或任意结构环境中失效，因此需要一种能处理此类环境并融合任务语义的鲁棒方法。

Result: 在多种模拟住宅数据集和大型开放式真实世界环境中验证，Prior-SG在语义区域分割准确率上达到SOTA，能在没有物理墙的情况下鲁棒划定远距离功能边界，并实现零样本本体论灵活性。

Insight: 创新点在于将场景图生成定义为概率对齐问题，并引入由大语言模型动态合成的先验图作为逻辑期望，通过马尔可夫随机场融合异构专家（视觉、几何、离散物体）与拓扑先验，以解决局部感知歧义并提供任务驱动的空间重组能力。

Abstract: Hierarchical 3D scene graphs are a promising representation for high-level spatial reasoning in autonomous mobile platforms. However, existing extraction frameworks typically rely on purely local visual clustering or strict geometric heuristics, such as wall-separated rooms, which fail in open-plan or arbitrarily-structured environments. We propose Prior-SG, a task- and prior-driven framework that casts scene graph generation fundamentally as a probabilistic alignment problem. As the robot explores, it continuously aggregates an incoming RGB-D sensor stream into a physically grounded Instance Graph utilizing a multi-scale, open-vocabulary feature fusion strategy. The system then infers the high-level functional semantics of this map through a Maximum A Posteriori (MAP) estimate, guided by a Prior Graph-a logical expectation of the environment’s structure and task-relevant vocabulary synthesized dynamically by a Large Language Model. By optimizing a Markov Random Field that fuses heterogeneous experts (visual, geometric, and discrete objects) with these topological priors, the system resolves local perceptual ambiguities. We validate this approach across diverse simulated residential datasets and large, open-plan real-world environments. Prior-SG achieves state-of-the-art semantic region segmentation accuracy compared to recent baselines, robustly delineates distant functional boundaries in the absence of physical walls, and uniquely provides zero-shot ontological flexibility, enabling the robot to entirely restructure its spatial partitioning based on a given high-level task.

Table of Contents

cs.CL [Back]

[1] Mean-Field Dynamics of Chain-of-Thought Reasoning in Large Language Models cs.CL | cs.AIPDF

[2] Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support cs.CL | cs.AIPDF

[3] RIG-RoPE: Relation- and Instance-Gated Rotary Positional Encoding with Duration-Aware Temporal Coordinates cs.CL | cs.CVPDF

[4] Cross-Architecture Steering Transfer in Language Models: A Systematic Empirical Study cs.CL | cs.LGPDF

[5] Conditional Cognitive Biases in LLMs: How Biased User Turns Modulate In-Context Reasoning cs.CL | cs.CY | cs.HC | cs.LGPDF

[6] ConWriter: Transition-Constrained Stateful Long-Form Story Generation with Lightweight Neuro-Symbolic Consistency Control cs.CLPDF

[7] Position: It’s Time to Optimize LLMs for Self-Consistency cs.CL | cs.AIPDF

[8] DREAM: LLM-based Dynamic Role-playing via Event-Aware Memory Graph cs.CL | cs.AIPDF

[9] Constraint-First Reasoning: A Training-Free Protocol for Exploiting Answer-Space Constraints in Mathematical Problem Solving cs.CL | cs.SCPDF

[10] Evidence Lock Before Commitment: A Frozen Interface Degrades LLM-as-Judge Evaluation cs.CLPDF

[11] Answer First, Reason Later: Commitment Order in Diffusion LLMs cs.CL | cs.AIPDF

[12] On-Policy Delta Distillation for Multilingual Math Reasoning cs.CL | cs.LGPDF

[13] Hierarchical Latent Prediction for Language Models cs.CL | cs.AIPDF

[14] M$^3$R-Bench: A Unified Benchmark for Evidence-Grounded Multimodal Metaphor Understanding cs.CLPDF

[15] MoCA: Implicit Social Context Analysis cs.CLPDF

[16] Enhancing Social Intelligence in LLMs with Hierarchical Reasoning and Utterance-Level Goal Rewarding cs.CLPDF

[17] MACRO: Markov Chain Routing of Transformer Layers cs.CL | cs.AIPDF

[18] Causal Episodic Memory for Feedback-Driven Agent Repair cs.CLPDF

[19] EpiBench: Can LLMs Understand Epitopes for Antibody Drug Discovery? cs.CL | q-bio.GNPDF

[20] Beyond Sequence Order: Syntax-Informed Positional Embeddings for Transformers cs.CL | cs.AIPDF

[21] NeSy-RAG: Neuro-Symbolic RAG for Explainable Question Answering cs.CL | cs.SCPDF

[22] Routing Is Least Learnable Where It Is Most Valuable: Bounds on Representation Routing for Web Agents cs.CLPDF

[23] RP-OPSD: Reasoning-Pivot-Guided On-Policy Self-Distillation for Multilingual Reasoning Transfer cs.CLPDF

[24] Learning When to Trust via Selective Context Preference Optimization cs.CL | cs.AI | cs.LGPDF

cs.CV [Back]

[25] MapTCL: Temporal Consistency Learning via Bidirectional Alignment for Vectorized HD Map Construction cs.CVPDF

[26] Innocent Panels, Hateful Stories: Evaluating and Detecting Hateful Intent in Multi-Turn Visual Story Generation cs.CV | cs.AI | cs.CRPDF

[27] Grad-CAM for Vision Transformers: A Systematic Taxonomy and Audit of Methodological Ambiguity in Explainable AI cs.CVPDF

[28] NeuroAdaptTrainer: A Fiji/ImageJ Plugin for YOLO-Based Neuron Segmentation, InteractiveCorrection and Transfer Learning cs.CVPDF

[29] In-Context Forcing: Uncovering Context Effects in Autoregressive Video Diffusion cs.CV | cs.AIPDF

[30] A Paragraph is Worth a Thousand Captions: Rethinking Text Supervision for Vision-Language Retrieval cs.CVPDF

[31] Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation cs.CV | cs.LGPDF

[32] LoDA: A Level of Detection Aware Method and a Multimodal Sensing Benchmark for Object Level Change Detection cs.CV | cs.ROPDF

[33] Adapting Vision Foundation Models with Cascaded Semantics cs.CVPDF

[34] Text-Guided Refinement of Multi-sequence Glioma Subregion Segmentation with a Vision-Language Foundation Model cs.CVPDF

[35] Invisible Shortcuts: Why Vision Encoders Know Your Camera cs.CV | cs.LGPDF

[36] VideoArgus: Agentic Rubric-Grounded Unified Evaluation for Video Generation and Editing cs.CVPDF

[37] DynaPix: Can Vision-Language Models Identify the Exact Future? cs.CVPDF

[38] APQF: Agentic Profiling-Guided Structured Pruning and Mixed-Precision Quantization with Adaptive Fine-Tuning cs.CV | cs.AI | cs.LGPDF

[39] HERA: Historical Evidence Routing Adapter for Physical Prediction in Latent World Models cs.CVPDF

[40] OmniMech: All-in-one Multimodal Mechanical Benchmark for 3D Reconstruction cs.CVPDF

[41] From Sports to Safety: Benchmarking Proactive Risk Inference in MLLMs cs.CV | cs.CLPDF

[42] EffectLearner: World-Aware Object-Effect Reasoning for Real-World Video Object Removal cs.CVPDF

[43] Beyond Frame Selection: Rethinking Long-Video Understanding with MLLMs cs.CVPDF

[44] TruthLens: Object Hallucination Detection via Self-Evaluating Truthfulness Scores in LVLMs cs.CVPDF

[45] SCI-CLIP: Segment-Centric Inference with Reference Memory for Training-Free Open-Vocabulary Segmentation cs.CVPDF

[46] ChronoVision: Temporal Reasoning via Latent State Reconstruction cs.CVPDF

[47] Vorch-Streamer: Extending Human Audio-Visual Generation to Real-Time Long-Form Streaming cs.CV | cs.SDPDF

[48] Vorch-IR: Long-Form Unified Multimodal Identity Replacement Video Generation cs.CVPDF

[49] DistMedVL: Distributional Vision-Language Alignment for Uncertainty-Aware Medical Image Segmentation cs.CV | cs.AI | cs.MMPDF

[50] URNet: A Unified Reparameterized Network for Efficient RGB-D Semantic Segmentation cs.CVPDF

[51] SciQNet: Two-Stage Multimodal Adaptation for Scientific Image Quality Assessment cs.CVPDF

[52] TAU-Bench: From Anomaly Instance Tracking to Fine-Grained Video Anomaly Understanding cs.CVPDF

[53] StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding cs.CVPDF

[54] G$^2$ARD-GS: Geometry-Guided Anchor-Regularized Gaussian Splatting Distillation cs.CVPDF

[55] LAWM-3D: Learning 3D-Aware Latent Actions from Human Videos for Generalizable Robot World Models cs.CVPDF

[56] One Ranking, Any Budget: Matryoshka Evidence-to-Context Frame Selection for Long-Video Understanding cs.CVPDF

[57] Engram-E2VID: Reference-Based Event-to-Video Reconstruction via Generative Activation of Appearance Engrams cs.CV | cs.LG | physics.opticsPDF

[58] Beyond Relevance: Bayesian Evidence Acquisition for Agentic Whole-Slide Image Reasoning cs.CVPDF

[59] UniVVT: A Unified End-to-End Framework for High-Fidelity Video Virtual Try-on cs.CV | cs.AIPDF

[60] GST-Bench: Can VLMs Develop Global Spatial Awareness from Video? cs.CVPDF

[61] Flow-Map Distillation on Relation Manifolds for Image Restoration cs.CVPDF

[62] HyTBE: Hyperbolic Target-Background Expert Model for Cross-Domain Infrared Small Target Detection cs.CV | cs.AIPDF

[63] Evidence-Driven Dynamic Visual Selector for Efficient Long Video Understanding cs.CVPDF

[64] Vorch-Director: Interactive World Story Model via Noise-Aware Error Rectification cs.CVPDF

[65] VSMP-IMU: Video-Grounded Semantic Motion Programs for Sensor-Aware Synthetic IMU Generation cs.CV | cs.LGPDF

[66] KVAE: Family of Tokenizers for Multimodal Generative Models cs.CV | cs.LG | cs.SDPDF

[67] Vorch-Omni: Multi-Task Orchestration of Sight and Sound cs.CVPDF

[68] Overcoming Attention Drift: Homogeneity-Heterogeneity Guided Feature Aggregation for Low-Light Remote Sensing Image Enhancement cs.CVPDF

[69] Curia-MAE: Multi-Modal Multi-Anatomy MAE Pre-Training for 3D Medical Image Segmentation cs.CVPDF

[70] D-CLOT: Double Closed Loop Optimal Transport for Unsupervised Action Segmentation cs.CV | cs.AIPDF

[71] Mapping Armenian Paris: Extracting and Geocoding Commercial Advertisements from the 20th-Century Diaspora Press cs.CVPDF

[72] Robust-WAM: Bridging Generative Pretraining and Semantic Foresight in World-Action Models cs.CV | cs.ROPDF

[73] Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models cs.CVPDF

[74] Big, Bright, or Invisible: A Frozen-Feature Benchmark of 3D CT Foundation Models cs.CV | cs.AIPDF

[75] Diff-VF: Training-free High-quality Long Video Generation via Diffusion Model cs.CVPDF

[76] Multi-Year Geospatial Reasoning using Interannually-Consistent Historical Predictions as a Free Input Modality cs.CVPDF

[77] Universal Concept Disruption for SAM3 Image Segmentation cs.CVPDF