cs.CL [Total: 25]
cs.CV [Total: 54]
eess.IV [Total: 1]
cs.LG [Total: 12]
cs.AI [Total: 8]
cs.MA [Total: 1]
cs.RO [Total: 2]
cs.IR [Total: 1]
hep-ex [Total: 1]
cs.HC [Total: 2]
cs.MM [Total: 1]

cs.CL [Back]

[1] Using Probabilistic Programs to Train Inductive Reasoning in Large Language Models cs.CL | cs.AI | cs.LG | stat.MLPDF

Liyi Zhang, Akshay K. Jagadish, Brenden M. Lake, Thomas L. Griffiths

TL;DR: 本文提出了一种名为基于程序的后验训练（PPT）的新方法，用于增强大语言模型（LLM）的归纳推理能力。该方法通过使用LLM生成多样化的开放世界场景作为概率程序，运行概率推断以产生对查询的分布性目标响应，并利用这些概率软标签进行微调。实验表明，PPT能显著提升模型在保留归纳任务上的估计准确性、与人类判断的一致性，并能迁移到外部基准测试中。

Details

Motivation: 当前针对LLM的后训练主要集中于可验证正确性的演绎任务（如数学和编码），而许多现实世界的推理问题是归纳性的，需要从稀疏、模糊的观察中推断不确定的信念。标准微调方法面临大规模高质量标注数据集的构建困难以及处理本质上是分布性目标的挑战。

Result: 在10,000个程序生成的场景上微调LLM后，评估显示PPT在保留的归纳任务上显著提高了估计准确性，增强了与人类判断的对齐，并能迁移到外部估计和校准基准测试中。此外，原始校准的增益无法通过后处理的温度缩放完全替代，表明模型更深入地内化了不确定性。

Insight: 创新点在于利用概率程序作为媒介来生成分布性训练数据（概率软标签），从而绕过高质量标注数据集的瓶颈，直接训练LLM处理不确定性并执行近似归纳推理。这为后训练LLM进行可靠的不确定性推理提供了一种有前景的途径。

Abstract: Post-training Large Language Models (LLMs) for reasoning typically focuses on deductive tasks such as mathematics and coding where correctness is verifiable. Yet, many real-world reasoning problems are inductive: agents must infer uncertain beliefs from sparse, ambiguous observations. There are challenges to using standard fine-tuning methods for inductive reasoning, including difficulties in curating large-scale, high-quality labeled datasets and in handling targets that are inherently distributional. In this work, we introduce a novel approach, called Program-based Posterior Training (PPT), to address these limitations: we use an LLM to generate diverse open-world scenarios as probabilistic programs, run probabilistic inference to produce distributional target responses to queries, and then fine-tune on these probabilistic soft labels. Using this approach, we fine-tune LLMs on 10,000 programmatically generated scenarios and evaluate on held-out motifs, human-labeled judgments, and external benchmarks. Overall, PPT substantially improves estimation accuracy on held-out inductive tasks, increases alignment with human judgments, and transfers to external benchmarks for estimation and calibration. Additionally, the gains in raw calibration are not subsumed by post-hoc temperature scaling, showing that the models have more deeply internalized uncertainty compared to output rescaling. Together, these results suggest that probabilistic-program-mediated fine-tuning is a promising approach for post-training LLMs to reliably perform approximate inductive inference.

[2] BenSyc: Benchmarking Conversational Sycophancy and Human Alignment in LLMs for Bengali Contexts cs.CLPDF

Kazi Noshin, Sajib Acharjee Dip, Ranat Das Prangon, Fardin Hassan Tamim, Syed Ishtiaque Ahmed

TL;DR: 本文介绍了BenSyc，这是首个针对孟加拉语社交语境中对话性谄媚行为（conversational sycophancy）的基准测试。该基准从孟加拉国和西孟加拉邦的Reddit社区收集了11,840个帖子和170k条评论，构建了一个包含二元标签和五级分类（从否定到升级）的人工验证数据集。研究评估了超过15个开源和专有大型语言模型（LLMs）在对话对齐分类和响应生成任务上的表现，发现即使是前沿的指令调优模型也难以区分共情支持与强化导向的认可，且在情感激烈情境下，多个模型频繁产生强烈认可或升级性回应。

Details

Motivation: 现有关于谄媚行为的研究主要集中于事实性同意和指令遵循场景，而基于文化的对话性谄媚尚未得到充分探索。LLMs越来越多地参与情感敏感的社交对话，其回应可能从平衡支持转向过度认可或升级性对齐，因此需要针对特定文化语境（如孟加拉语）的基准来评估社交对齐的对话AI系统。

Result: 在分类任务中，最佳系统在二元检测上的宏平均F1分数仅为61.8，在五类分类上为61.7。在生成任务中，多个模型在情感激烈情境下频繁产生强烈认可或升级性回应。结果表明，区分共情支持与强化导向的认可对前沿指令调优模型仍具挑战性。

Insight: 论文的创新点在于构建了首个针对孟加拉语社交语境中对话性谄媚行为的基准测试BenSyc，并引入了细粒度的五级分类法。从客观角度看，该研究强调了基于文化的多语言基准对于评估社交对齐对话AI系统的重要性，揭示了不同模型系列和对话行为间的显著差异，为LLMs在情感敏感对话中的对齐研究提供了新的评估维度。

Abstract: Large language models (LLMs) increasingly participate in emotionally sensitive social conversations, where responses may shift from balanced support toward excessive validation or escalatory alignment. Existing sycophancy research primarily focuses on factual agreement and instruction-following settings, leaving culturally grounded conversational sycophancy underexplored. We introduce BenSyc, the first benchmark for studying conversational sycophancy in Bengali social contexts. Starting from 11,840 Reddit posts and 170k comments collected from communities across Bangladesh and West Bengal, we construct a human-validated benchmark with binary labels and a fine-grained five-level taxonomy spanning Invalidation, Neutral, Support, Validation, and Escalation. We evaluate more than 15 open and proprietary LLMs on conversational alignment classification and response generation tasks. Results show that distinguishing empathetic support from reinforcement-oriented validation remains challenging even for frontier instruction-tuned models: the best system achieves only 61.8 Macro-F1 on binary detection and 61.7 Macro-F1 on five-class classification. In generation settings, several models frequently produce strongly validating or escalatory responses in emotionally charged situations. Our findings highlight substantial variation across model families and conversational behaviors, underscoring the importance of culturally grounded multilingual benchmarks for evaluating socially aligned conversational AI systems.

[3] CodeAlchemy: Synthetic Code Rewriting at Scale cs.CL | cs.LGPDF

Ankit Gupta, Aditya Prasad, Rameswar Panda

TL;DR: CodeAlchemy是一个大规模合成代码重写框架，通过五种策略将公开代码转化为语义丰富的训练数据：质量感知重写、模板化问题、开发者任务、多轮对话和执行追踪。该框架处理了15种语言的3个语料库，生成超过500B标记的合成数据和350B推理标记，并创建了DevEval和TraceEval两个新基准。其3B参数模型在多个代码生成基准上超越了规模大10倍的前沿模型。

Details

Motivation: 现有代码预训练主要学习语法，但对多样化真实世界任务格式的信号稀疏；合成数据虽在语言模型中有效，但在代码领域尚未充分探索，仅限于有限的质量改进。

Result: 在HumanEval上达到83.5%，MBPP上63.2%，DevEval上8.09%胜率，TraceEval上15.36 ROUGE-2分数，超越了包括27B Gemma-3和32B Granite-4.0在内的规模大10倍的前沿模型；新基准TraceEval上Claude Sonnet 4.5仅5.6%精确匹配，揭示了语义理解的重大差距。

Insight: 创新点在于系统性地通过五种策略生成大规模、语义丰富的合成代码数据，特别是CodeTrace通过执行1.3M+文件捕获控制流、状态跟踪和库知识；客观分析认为，将代码执行追踪纳入训练数据是提升模型语义理解的关键突破，且小模型通过高质量合成数据可超越大模型。

Abstract: Pre-training on raw code teaches syntax but provides sparse signal for diverse real-world task formats. While synthetic data has proven transformative for language models, code remains largely unexplored beyond limited quality improvements. We present CodeAlchemy, a synthetic data generation framework that transforms publicly sourced code into semantically-rich training data through 5 strategies: CodeEnhance (quality-aware rewriting), CodeQA (template-based problems), CodeDev (developer tasks), CodeDialogue (multi-turn conversations), and CodeTrace (execution traces). We process 3 corpora across 15 languages to generate 500B+ tokens of synthetic data plus 350B reasoning tokens, orders of magnitude more than prior efforts. CodeTrace instruments and executes 1.3M+ files across 14 languages and 5K libraries, capturing control flow, state tracking, and library knowledge. We introduce DevEval (developer tasks) and TraceEval (execution prediction) benchmarks; frontier models like Claude Sonnet 4.5 achieve only 5.6% exact match on TraceEval, revealing critical gaps in semantic understanding. Our 3B models achieve 83.5% on HumanEval, 63.2% on MBPP, 8.09% win rate on DevEval, and 15.36 ROUGE-2 on TraceEval, outperforming frontier models 10x the size including 27B Gemma-3 and 32B Granite-4.0.

[4] Pareto-Guided Teacher Alignment for Fair Personalized Text Generation cs.CL | cs.AI | cs.CY | cs.LGPDF

Tunazzina Islam

TL;DR: 本文研究个性化说服文本生成中的公平性问题，提出将公平性缓解视为约束多目标对齐问题：在减少人口统计差异的同时保持个性化保真度。作者开发了帕累托引导的教师对齐框架，包含基于修订的候选生成、配对感知可行性门控、帕累托式候选选择以及可选的有监督微调和直接偏好优化。在气候变化和疫苗接种说服任务上，使用包含匹配性别和年龄对的受控上下文丰富人口统计网格以及统一的五审计评估套件进行评估。

Details

Motivation: 个性化说服文本生成虽然能提高相关性和参与度，但基于人口统计的调节可能在不同群体间引入不平等的框架。本文旨在解决个性化生成中的公平性缓解问题，将其构建为一个约束多目标对齐问题。

Result: 评估结果显示，在跨领域和跨模型族迁移设置中，没有任何一种对齐策略能在所有目标上同时占优。不同方法占据了公平性-个性化帕累托前沿的不同区域：一些方法实现了更强的差异减少，而另一些则更好地保持了个性化或人口统计稳定性。公平性缓解效果依赖于具体目标，并且在跨领域和模型族之间转移不一致。

Insight: 论文的创新点在于将公平性缓解形式化为多目标对齐问题，并提出了帕累托引导的教师对齐框架。从客观角度看，其核心洞察是：对于公平敏感的个性化生成，应采用有界回归、多审计模型选择，而非单一指标优化，因为方法在公平性与个性化之间存在权衡，形成了帕累托前沿。

Abstract: Personalized persuasive text generation can improve relevance and engagement, but demographic conditioning may also introduce unequal framing across groups. We study fairness mitigation in personalized generation as a constrained multi-objective alignment problem: reduce demographic disparities while preserving personalization fidelity. We propose a Pareto-guided teacher alignment framework that combines revision-based candidate generation, pair-aware feasibility gating, Pareto-style candidate selection, and optional preference optimization through supervised fine-tuning and direct preference optimization. We evaluate the framework on climate change and vaccination persuasion tasks using a controlled context-rich demographic grid with matched gender and age pairs and a unified five-audit evaluation suite spanning persuasion bias, formality disparity, emotional framing disparity, lexical association disparity, and personalization fidelity. Across both domains and cross-family transfer settings, no single alignment strategy dominates all objectives simultaneously. Instead, methods occupy different regions of a fairness-personalization Pareto frontier: some achieve stronger disparity reductions, while others better preserve personalization or demographic stability. Our results show that fairness mitigation effects are objective-dependent and transfer inconsistently across domains and model families, motivating bounded-regression, multi-audit model selection over single-metric optimization for fairness-sensitive personalized generation.

[5] OpenRTLSet: A Fully Open-Source Dataset for Large Language Model-based Verilog Module Design cs.CLPDF

Jinghua Wang, Lily Jiaxin Wan, Sanjana Pingali, Scott Smith, Manvi Jha

TL;DR: OpenRTLSet是一个用于基于大语言模型的Verilog模块设计的完全开源数据集，提供了超过13.1万个多样化的Verilog代码样本。该数据集独特地整合了来自GitHub仓库的Verilog代码、VHDL翻译以及可综合的C/C++翻译，并为每个样本生成了配对的自然语言描述，可用于微调各种语言模型以进行Verilog代码生成。

Details

Motivation: 旨在解决硬件设计领域缺乏大规模、完全开源数据集的问题，为研究和工业界提供一个无障碍的基准，以探索和开发基于LLM的硬件设计工具。

Result: 论文展示了开源方法在硬件设计任务中可以实现优越性能，通过探索不同选项（如Verilator生成的C++文件作为额外上下文、量化技术、不同模型规模）为领域研究建立了新的基础。

Insight: 主要创新点在于构建了首个大规模、完全开源的硬件设计数据集，并利用推理模型DeepSeek-R1自动生成代码-描述对。其数据集的多源构成（GitHub、翻译代码）和实验中对不同技术选项的系统性探索，为开源硬件设计LLM的研究提供了宝贵的资源和基准。

Abstract: OpenRTLSet introduces the largest fully open-source dataset for hardware design, offering over 131,000 diverse Verilog code samples to the research community and industry. Our dataset uniquely combines Verilog code from GitHub repositories (102k modules), VHDL translations (5k modules), and synthesizable C/C++ translations (24k modules), all freely accessible without proprietary restrictions. Using the reasoning model DeepSeek-R1, we generated paired natural language descriptions for each code sample, enabling fine-tuning of various language model families (e.g., Qwen and Granite) for Verilog code generation. Our dataset explores multiple options, including Verilator-generated C++ files as additional context during labeling, quantization techniques (INT4 vs. BF16), and performance differences across model sizes (7B-32B parameters). OpenRTLSet demonstrates that open-source approaches can achieve superior performance in hardware design tasks, establishing a new foundation for accessible research and commercial use in this domain.

[6] The Confident Liar: Diagnosing Multi-Agent Debate with Log-Probabilities and LLM-as-Judge cs.CL | cs.AIPDF

Ali Keramati, Justin Cheok, Jacob Horne, Mark Warschauer

TL;DR: 本文研究了多智能体辩论系统中内部置信度信号（如词元对数概率）与外部评估的推理质量（如LLM-as-judge评分）及任务准确性之间的关系。通过一个包含构造者和审计者的两智能体辩论架构，结合LLM-as-judge对推理过程进行多维度评分，论文在基于评分准则、数学推理和事实问答三个领域进行了实验分析。

Details

Motivation: 现有对多智能体辩论系统的评估通常只关注最终答案的正确性，而忽视了辩论旨在产生的中间推理过程的质量。本文旨在探究内部置信度信号是否能预测外部评估的推理质量，以及这些信号是否与任务正确性一致。

Result: 在基于评分准则的领域实验中，发现置信度轨迹呈现一致的四阶段模式，且构造者与审计者之间存在显著的角色不对称性：构造者的置信度与评估的推理质量的一致性大约是审计者的两倍，且构造者基于置信度检测关键推理失败的可靠性（AUROC 0.804）明显高于审计者（0.634）。

Insight: 创新点在于提出了一个结合词元级对数概率分布、LLM-as-judge评分和任务准确性的多信号分析框架，用于诊断多智能体辩论的推理质量。客观来看，该研究揭示了辩论智能体角色间的置信度不对称性，为优化辩论系统设计提供了新的诊断视角。

Abstract: Multi-agent debate systems are typically evaluated only on whether the final answer is correct, overlooking the quality of the intermediate reasoning that debate is designed to produce. This paper studies the relationship between three signals in multi-agent debate: token-level log-probability distributions over reasoning tokens, LLM-as-judge rubric scores assigned to those tokens, and final task accuracy. We examine whether internal confidence signals predict externally evaluated reasoning quality, and whether either signal aligns with task correctness, across three domains: rubric-based scoring, mathematical reasoning, and factual question answering. Our framework pairs a two-agent debate architecture – a Constructor and an Auditor – with an LLM-as-judge that scores each agent’s reasoning along instruction following, justification quality, and evidence grounding, together with a critical-failure flag. Experiments in the rubric-scoring domain reveal a consistent four-phase confidence trajectory and a substantial role asymmetry: confidence aligns with judged reasoning quality roughly twice as strongly for the Constructor as for the Auditor, and confidence-based detection of critical reasoning failures is markedly more reliable for the Constructor (AUROC 0.804) than for the Auditor (0.634). These findings motivate the broader cross-domain investigation proposed in this paper.

[7] Early-Token Confidence Predicts Reasoning Quality in Multi-Agent LLM Debate cs.CLPDF

Ali Keramati, Justin Cheok, Jacob Horne, Mark Warschauer

TL;DR: 本文研究了在多智能体LLM辩论系统中，利用解码过程中的词元级对数概率作为内在置信度信号来预测推理质量。通过基于辩论的论文评分框架，在ASAP数据集上比较了置信度代理与基于评分标准的评委分数，发现早期词元置信度（尤其是生成的前几个词元）是推理质量的最强预测指标，优于全序列统计量。

Details

Motivation: 在多智能体LLM系统中，评估开放端任务（无参考答案）的推理质量具有挑战性，因此探索能否利用解码过程中的内在置信度信号（如词元级对数概率）来预测LLM作为评委评估的推理质量。

Result: 在ASAP论文数据集上的实验表明，早期词元置信度是推理质量的最强预测指标，优于全序列统计量；分析还发现支持性推理的置信度与质量对齐度高于对抗性批判，存在系统性的角色不对称性。

Insight: 创新点在于揭示了生成过程的早期阶段（前几个词元）的置信度信号最具信息量，为多智能体LLM系统提供了一种轻量且有效的推理可靠性估计方法；客观分析认为，对解码动态轨迹的深入分析为理解LLM推理过程提供了新视角。

Abstract: Evaluating reasoning quality in multi-agent LLM systems is challenging, especially for open-ended tasks without reference answers. We investigate whether intrinsic confidence signals, token-level log-probabilities from decoding, can predict reasoning quality as assessed by LLM-as-judge evaluation. Using a debate-based essay scoring framework, we compare confidence proxies against rubric-based judge scores across two ASAP essay sets. We find that early-token confidence, particularly within the first few generated tokens, is consistently the strongest predictor of reasoning quality, outperforming full-sequence statistics. Analysis of log-probability trajectories shows that the opening phase of generation is the most heterogeneous and therefore most informative. We also observe a systematic asymmetry between agent roles, with stronger alignment between confidence and quality for supportive reasoning than for adversarial critique. These results suggest that early decoding dynamics provide a lightweight and effective signal for estimating reasoning reliability in multi-agent LLM systems.

[8] TabClaw: An Interactive and Self-Evolving Agent for Spreadsheet Manipulation and Table Reasoning cs.CLPDF

Mingyue Cheng, Shuo Yu, Daoyu Wang, Qingchuan Li, Xiaoyu Tao

TL;DR: TabClaw是一个开源的交互式AI智能体，用于电子表格操作和表格推理。用户上传CSV或Excel文件并用自然语言发出请求，TabClaw会澄清模糊意图、展示可编辑的执行计划、流式执行ReAct风格的工具使用分析循环、调度专家智能体进行并行多表推理，并综合带有明确共识和不确定性标记的发现。除了单次分析，它还能记录工作流、提取持久用户记忆、从重复工具使用模式中提炼可重用技能、支持包式技能导入，并通过负面反馈升级技能。

Details

Motivation: 解决现有LLM智能体在电子表格和表格数据分析中透明度有限、依赖隐含假设、难以进行多表比较以及无法适应用户偏好和重复工作流的问题。

Result: 在电子表格操作和表格推理基准测试上的实验表明，TabClaw提高了可执行任务的完成度和推理性能，同时保持了可检查的用户工作流。

Insight: 创新点在于将交互性、可检查性、工作流记录与个性化学习（包括用户记忆提取、技能提炼与升级）结合到一个统一的智能体框架中，实现了从单次分析到持续自我演化的转变，提升了透明度和适应性。

Abstract: Spreadsheets and tables are widely used representations for structured data analysis, but effective analysis still requires substantial manual effort and domain expertise. Recent large language model (LLM) agents can automate parts of this process, but they often provide limited transparency into intermediate decisions, rely on implicit assumptions, struggle with multi-table comparison, and repeat similar workflows without adapting to a user’s preferences. This paper presents TabClaw, an open-source interactive AI agent for spreadsheet manipulation and table reasoning. Users upload CSV or Excel files and issue natural-language requests; TabClaw clarifies ambiguous intent, exposes an editable execution plan, streams a ReAct-style tool-using analysis loop, dispatches specialist agents for parallel multi-table reasoning, and synthesizes findings with explicit consensus and uncertainty markers. Beyond one-off analysis, TabClaw records completed workflows, extracts persistent user memory, distills reusable skills from repeated tool-use patterns, supports package-style skill import, and upgrades skills from negative feedback. Experiments on spreadsheet manipulation and table reasoning benchmarks show that TabClaw improves executable task completion and reasoning performance while preserving an inspectable user workflow. This paper shows how TabClaw turns spreadsheets and tables into inspectable analytical workflows while gradually personalizing itself to recurring data-analysis tasks. Our code is available.

[9] PADD: Path-Aligned Decompression Distillation for Non-Router Teacher to Guide MoE Student Learning cs.CL | cs.LGPDF

Xinyue Peng, Yi Qian, Jiaojiao Lin, Wenjian Shao, Yanming Liu

TL;DR: 本文提出了一种名为路径对齐解压缩蒸馏（PADD）的框架，用于将无显式路由的稠密教师模型的知识蒸馏到混合专家（MoE）学生模型中，并同时学习高质量的路由策略。该方法通过两阶段（初始化与训练）四步骤的流程，包括教师神经元聚类、学生专家预热、在线自适应蒸馏、路径精炼策略优化和奖励增强负载均衡。在数学推理基准测试上的实验表明，该方法在相同推理成本下显著优于强基线，且MoE学生模型能匹配甚至超越其稠密教师模型。

Details

Motivation: 随着大语言模型规模持续扩大，在固定计算预算下提升模型容量变得日益困难。本文旨在解决如何将稠密教师模型的知识有效蒸馏到MoE学生模型中，并同时学习高质量路由策略的问题。

Result: 在数学推理基准测试（如GSM8K和MATH）上，PADD方法在相同推理成本下相比强基线（如标准知识蒸馏和MoE训练方法）取得了显著提升，其MoE学生模型能够匹配甚至超越原始稠密教师模型的性能，同时展示了稳定的路由行为。

Insight: 主要创新点在于将知识蒸馏与路由策略学习统一到一个四阶段训练流程中，特别是通过教师神经元聚类初始化学生专家功能多样性，以及结合在线自适应蒸馏和路径精炼策略优化来协同优化模型参数与路由策略。从客观角度看，其将蒸馏过程解耦为功能初始化与联合训练的思路，为高效训练MoE模型提供了可借鉴的框架。

Abstract: As large language models (LLMs) continue to scale, it becomes increasingly challenging to grow model capacity under fixed computation budgets. We propose Path-Aligned Decompression Distillation (PADD), a framework for distilling knowledge from dense teachers without explicit routing into mixture-of-experts (MoE) students while learning high-quality routing policies. PADD organizes knowledge distillation into four stages in two phases: an initialization phase (Stage I) that builds diverse functionality in the student’s experts through teacher neuron clustering and student-expert warmup, and a training phase (Stages II–IV) that integrates online adaptive distillation, path-refined policy optimization, and reward-augmented load balancing in a single training pipeline. Experiments on mathematical reasoning benchmarks demonstrate that PADD yields substantial gains over strong baselines at the same inference cost and that the MoE student can match or surpass its dense teacher. They also demonstrate effective teacher-to-student knowledge distillation and stable routing behavior.

[10] Do Vision-Language Models See or Guess? Measuring and Reducing Textual-Prior Reliance with a Phrasing-Controlled Benchmark cs.CL | cs.CVPDF

Pratham Singla, Shivank Garg, Vihan Singh, Paras Chopra

TL;DR: 本文构建了一个名为’短语控制基准’的540张图像数据集，用于测量视觉语言模型对文本先验的依赖程度。研究发现，所有测试的11个VLMs在难度最高的变体问题上性能均下降，且开源模型退化最严重。通过无图像消融实验等方法证实了模型确实存在对图像的依赖不足问题，并展示了通过上下文示例和GRPO后训练可以有效缓解此问题。

Details

Motivation: 视觉语言模型在部署时，其答案应基于图像内容，但它们往往过度依赖问题措辞和记忆中的世界知识（即文本先验），而非图像本身，这导致基准分数虚高并产生自信但缺乏依据的答案。现有基准难以孤立测量这种行为。

Result: 在构建的六类推理任务基准上，所有11个模型（从开源小模型到闭源大系统）在直接根据图像生成、文本信息泄露最少的最难变体问题上性能均下降。无图像消融实验显示，开源模型性能跌至纯文本基线水平（1%到9%）。通过上下文示例匹配和GRPO后训练可以提升模型在所有变体上的准确率，并能泛化到分布外数据集。

Insight: 创新点在于提出了一个通过控制问题措辞变量来量化文本先验依赖的基准构建方法。客观分析认为，其核心诊断方法（无图像消融）和提出的缓解策略（上下文示例和GRPO训练）为理解和改进VLMs的视觉基础能力提供了可复现的测量框架和可行路径。

Abstract: Vision-language models (VLMs) are increasingly deployed where answers must follow from what is in the image, yet they often answer from textual priors, the question’s phrasing together with memorized world knowledge, rather than from the image itself, which inflates benchmark scores and yields confident but ungrounded answers. Existing benchmarks rarely isolate this behavior, since each image is usually paired with a single fixed question. To measure the reliance, we build a 540-image benchmark across six reasoning categories and generate four question variants over the same images, so that phrasing rather than image content is the controlled variable. The hardest variant is written directly from the image to minimize text leakage. We benchmark eleven VLMs spanning small open-weight models to large closed-source systems: every model degrades on the hardest variant, and open models fall furthest. Our central diagnostic is a no-image ablation, which collapses the open-weight models to their text-only floor (1 to 9 percent). Three further analyses, LLM-rated difficulty, low base-to-final textual similarity, and human re-annotation, corroborate genuine image-dependence. In-context exemplars that match how a variant was built recover the most accuracy, and GRPO post-training of a small VLM yields consistent gains across all four variants that transfer to a held-out out-of-distribution set. Textual-prior reliance is measurable and partly trainable away.

[11] LakeQA: An Exploratory QA Benchmark over a Million-Scale Data Lake cs.CL | cs.AIPDF

Haonan Wang, Jiaxiang Liu, Yurong Liu, Austin Senna Wijaya, Tianle Zhou

TL;DR: 该论文提出了LakeQA，一个面向数据湖的搜索式问答基准，旨在评估大语言模型在需要搜索和推理的复杂问答任务上的能力。该基准基于约9.5TB的异构文本资源构建，包含维基百科和政府开放数据，任务需要多跳推理和跨源证据整合。实验表明，即使是前沿LLM在该基准上表现也极具挑战性。

Details

Motivation: 现有LLM在阅读式问答上进展迅速，但现实世界的问题往往没有明确的证据文档，需要从海量数据湖中搜索。目前缺乏一个要求在大规模数据湖上进行搜索和推理的综合基准。

Result: 在七个前沿大语言模型上的实验结果表明，LakeQA具有挑战性。例如，GPT-5.2在LakeQA上的精确匹配分数仅为18.37%。

Insight: 论文的创新点在于构建了一个大规模、异构、需要长程多跳推理的搜索式问答基准，填补了该领域的空白。其核心价值在于提供了一个更贴近现实、能同时评估搜索和推理能力的测试平台，对开发能在现代数据湖中查找和分析数据的LLM智能体至关重要。

Abstract: Recent large language models (LLMs) have shown rapid progress in reading-based question answering (QA), where evidence is explicitly provided or can be trivially retrieved. In contrast, real-world questions are often not paired with accurate evidence documents. The useful evidence resides in massive data lakes, making search a prerequisite for answering. However, there is a lack of comprehensive benchmarks that require both searching and reasoning over large data lakes. To this end, we introduce LakeQA, a comprehensive benchmark for search-centric question answering over data lakes that jointly emphasizes searching and reasoning capabilities. LakeQA is built on a heterogeneous collection of approximately 9.5 TB of text resources from Wikipedia and open-source government data, spanning structured and unstructured data. To ensure task quality, each sample is annotated by at least one Ph.D.-level expert. Each task requires long-horizon multi-hop reasoning with implicit intermediate steps: agents need to discover the correct documents and then compose evidence across sources to produce the answer. Experimental results on seven frontier LLMs demonstrate that LakeQA is challenging. For instance, GPT-5.2 achieves only an exact-match score of 18.37% on LakeQA. Overall, LakeQA provides a realistic testbed for developing LLM agents that can both find and analyze data in modern data lakes.

[12] KCSAT-ML: Probing Reasoning Models with Nationwide-Cohort Human Difficulty cs.CLPDF

Sanghee Park, Geewook Kim, Kee-Eung Kim

TL;DR: 本文介绍了KCSAT-ML数据集，这是一个基于韩国高考（KCSAT）数学试题构建的推理基准，包含664道题目，其中339道核心题目附有基于数十万考生真实表现的官方错误率。论文提出了难度对齐推理增益（DRG）指标，用于评估模型错误是否与人类认为的题目难度一致。研究发现，在不同规模的视觉语言模型（VLM）和通过OCR处理的LLM中，存在预算限制下高难度题目准确率骤降、测试时扩展（TTS）导致计算开销与错误率线性增长但准确率提升非单调、以及同一模型系列在难易题目上出现反扩展和过度思考等对齐失败模式。

Details

Motivation: 现有数学推理基准大多缺乏基于真实人类表现的逐题难度信号，无法精细评估模型与人类推理难度的对齐程度。

Result: 在KCSAT-ML基准上，研究发现低预算模型的准确率在人类高错误率题目上显著下降；测试时扩展（TTS）使计算量随人类错误率线性增加，但准确率提升呈非单调曲线；DRG指标揭示，即使准确率相近的模型，其错误模式也可能截然不同（一个在人类认为难的题目上出错，另一个则在人类认为简单的题目上失败）。

Insight: 创新点在于引入了基于大规模人类表现数据的逐题难度标注数据集（KCSAT-ML）和难度对齐推理增益（DRG）评估指标，能够揭示模型错误模式与人类认知难度的一致性，这是传统聚合准确率指标所掩盖的。从客观角度看，该研究为模型推理能力的细粒度评估和对齐失败分析提供了新工具和新视角。

Abstract: Math reasoning benchmarks have proliferated, yet most lack a per-item difficulty signal grounded in actual human performance. We introduce KCSAT-ML, a decade (2014-2025) of Korean College Scholastic Ability Test (KCSAT; Suneung) mathematics: 664 problems with a 339-item core set carrying official per-item error rates from nationwide cohorts of hundreds of thousands of examinees. We pair the benchmark with Difficulty-aligned Reasoning Gain (DRG): a score-orthogonal metric that asks whether a model’s mistakes concentrate on the items humans found hard, or on items humans found easy. Together they expose, across a wide range of VLMs (and LLMs via OCR), three patterns: (i) low-budget accuracy collapses on the high-human-error tail at every model size; (ii) test-time scaling (TTS) raises token use roughly linearly with cohort error rate, while accuracy gains follow a non-monotonic curve; (iii) within a single family, TTS flips between anti-scaling on the hardest items and overthinking on easier ones – two faces of the same alignment failure. On DRG, models with near-identical accuracy can sit at near-opposite values: one model gets wrong what humans also find hard, while another solves the hardest items yet fails on items humans find easy – a contrast that aggregate accuracy hides. Our code and dataset builder will be open-sourced at https://github.com/naver-ai/KCSAT-ML.

[13] WebChallenger: A Reliable and Efficient Generalist Web Agent cs.CLPDF

Jayoo Hwang, Xiaowen Zhang, Vedant Padwal

TL;DR: 本文提出了WebChallenger，一个高效、通用的网页智能体框架。其核心是PageMem，一种从DOM确定性构建的结构化页面表示，将页面呈现为带有简短摘要的语义区域层次结构。基于此，框架通过三个机制模仿人类认知优势：选择性注意、持久记忆和流程化交互，从而在不依赖大规模专有模型或网站特定适配器的情况下，实现了跨网站的通用网页导航。

Details

Motivation: 当前LLM智能体在自主网页导航任务中面临挑战，最强的通用系统依赖昂贵的专有推理模型，这限制了其在重复性任务中的应用。作者认为问题不在于模型能力不足，而在于现有智能体架构未能复制人类在网页交互中的三个认知优势：对相关区域的选择性注意、对网站结构的持久记忆以及对常见交互模式的流程化操作。

Result: 在未对模型进行微调的情况下，使用现成的开源模型，WebChallenger在多个基准测试中取得了接近前沿专有系统的性能：WebArena上达到56.3%，VisualWebArena上达到48.7%，Online-Mind2Web上达到51.0%，WorkArena上达到70.9%。

Insight: 主要创新点在于通过架构设计而非模型规模来提升性能。其核心是PageMem这一结构化页面表示，它作为共享基础，支撑了三个模仿人类认知的机制。这种设计使得框架具有跨网站的通用性，无需为每个网站定制适配器，同时显著降低了推理成本，为构建高效、可靠的通用网页智能体提供了新思路。

Abstract: Autonomous web navigation remains challenging for LLM agents, and the strongest generalist systems rely on proprietary reasoning models whose inference cost is prohibitive for the repetitive tasks where such agents would be most useful. We argue this gap stems not from insufficient model capability but from agent architectures that fail to replicate three human cognitive advantages: selective attention to relevant page regions, persistent memory of website structure, and procedural fluency with common interaction patterns. We introduce WebChallenger, a web agent framework that addresses each gap through architecture design rather than model scale, built around PageMem: a structured page representation deterministically constructed from the DOM that exposes each page as a hierarchy of semantic sections with short summaries. On this shared substrate we build three mechanisms that mirror the three cognitive advantages: a divide-and-conquer observation pipeline that lets the agent skim section summaries and extract details only from task-relevant regions; a lightweight exploration and memory system that traverses each website once to build a reusable map of pages and element behaviors; and compound action workflows that collapse common multi-step interactions into single agent actions, handling partial state changes automatically. Because all three operate over PageMem, the framework generalizes across websites without site-specific adapters. Using off-the-shelf open-weight models without fine-tuning, our system achieves 56.3% on WebArena, 48.7% on VisualWebArena, 51.0% on Online-Mind2Web, and 70.9% on WorkArena, approaching frontier proprietary systems at a fraction of the cost. Our code is released at https://github.com/jayoohwang1/webchallenger

[14] ParaBridge: Bridging Paralinguistic Perception and Dialogue Behavior in Speech Language Models cs.CL | cs.SD | eess.ASPDF

Yuxiang Wang, Qinke Ni, Shengbo Cai, Wan Lin, Liqiang Zhang

TL;DR: 本文提出了一种名为ParaBridge的在线策略自蒸馏方法，旨在弥合语音语言模型在感知副语言线索（如语调、背景噪音）与生成对话行为之间的差距。该方法通过将推理阶段脆弱的副语言指令支架转化为稳定的模型行为，无需人工标注或外部奖励模型，即可教导模型何时以及如何根据非词汇线索调整回复。

Details

Motivation: 当前语音语言模型能够识别副语言线索，但在开放域对话中常常忽略这些信息，导致感知与行为之间存在差距。现有的推理阶段指令支架方法在多轮上下文和竞争指令下表现脆弱，需要一种更稳定的训练方法。

Result: 在Qwen3-Omni-thinking模型上，ParaBridge将无支架的VoxSafeBench SAR从14.6%提升至40.3%，并将EchoMind平均评分从3.27提高到3.92。同时，模型在MMAU-Pro、VoiceBench和GPQA等基准测试中保持了通用能力，与原模型差距在0.4分以内。

Insight: 创新点在于提出了一种自蒸馏框架，将推理时脆弱的指令支架作为训练时的特权视图，为无支架模型提供密集的全词汇下一个token监督，从而稳定地整合副语言感知。该方法能够泛化到未见过的副语言线索，从安全导向训练迁移到共情导向对话，并适用于不同的SLM骨干网络。

Abstract: Speech carries more information than just words: a child’s voice, a fearful tone, or a noisy background should all lead a sufficiently competent spoken-dialogue assistant to different replies. Current Speech Language Models (SLMs) can recognize such paralinguistic cues but often ignore them in open-ended dialogue. We observe that a simple paralinguistic instruction scaffold at the inference stage narrows this perception-behavior gap, suggesting that the relevant cues are already latent in the model. Such scaffolds, however, remain brittle under multi-turn context and competing instructions. Therefore, we propose \textbf{ParaBridge}, an on-policy self-distillation method that turns a brittle inference-time scaffold into stable model behavior. During training, the scaffold serves only as a temporary privileged view; the scaffold-free model rolls out its own response, while the scaffolded view supplies dense, full-vocabulary next-token targets along its trajectory. This supervision teaches when non-lexical cues should affect the reply without the need for curated dialogues, human labels, or external reward models. On Qwen3-Omni-thinking, ParaBridge raises scaffold-free VoxSafeBench SAR from $14.6%$ to $40.3%$ and improves EchoMind average rating from $3.27$ to $3.92$. It also preserves general ability, with MMAU-Pro, VoiceBench, and GPQA all within $0.4$ points of the original model. Beyond the training distribution, ParaBridge generalizes to unseen paralinguistic cues, transfers from safety-oriented training to empathy-oriented dialogue, and works on a different SLM backbone.

[15] REAL: A Reasoning-Enhanced Graph Framework for Long-Term Memory Management of LLMs cs.CLPDF

Keer Lu, Liwei Chen, Guoqing Jiang, Zhiheng Qin, Yunhuai Liu

TL;DR: 本文提出了REAL，一个用于大语言模型长期记忆管理的推理增强图框架。它通过构建一个时序和置信度感知的有向属性图来存储历史交互信息，并采用非破坏性更新策略来追踪事实演变，同时结合语义评估器引导的混合束搜索和反事实推理来增强记忆检索的准确性和完整性。

Details

Motivation: 解决LLMs因有限上下文窗口而无法长期保留所有历史交互信息的问题，以及现有外部记忆系统存在的三大局限：基于扁平文本的组织方式无法捕捉记忆间显式关系、结构化记忆系统常破坏性地覆盖演变事实、现有检索机制在证据不完整时仍保持查询无关和被动。

Result: 综合实验表明，REAL在长期记忆性能上显著优于基于扁平文本、基于图以及现有的记忆基线方法，平均提升了22.72%。

Insight: 创新点在于将长期对话记忆建模为时序和置信度感知的有向属性图，并提出了非破坏性时序更新策略以保留并行事实版本及其有效区间，以及在检索阶段通过锚定根实体、解耦探索意图、进行语义评估器引导的混合束搜索并结合反事实推理来修复不可靠状态和恢复缺失证据，从而实现对事实演变的忠实追踪和更精准、主动的检索。

Abstract: Large Language Models (LLMs) are increasingly expected to interact with users over long time horizons. However, due to their finite context window, LLMs cannot retain all past interactions, making long-term memory management essential for storing, updating, and retrieving historical information beyond the context limit. Although recent memory systems attempt to address this issue by storing historical information externally, existing approaches suffer from three key limitations: flat text-based memory organizations fail to capture explicit relations among memories, structured memory systems often destructively overwrite evolving facts, and current retrieval mechanisms remain query-agnostic and passive when evidence is incomplete. REAL constructs long-term conversational memory as a temporal and confidence-aware directed property graph, where each atomic fact is represented with entities, relations, valid-time intervals, confidence scores, and exploration intent labels. During memory construction, REAL adopts a non-destructive temporal update strategy that preserves parallel fact versions and their validity intervals, enabling faithful tracking of fact evolution. During retrieval, REAL anchors query-relevant root entities, decouples their exploration intents, and performs semantic evaluator-guided hybrid beam search to extract compact memory subgraphs. It further incorporates counterfactual inference to repair unreliable retrieval states and recover missing memory evidence through implicit logical relations. Comprehensive experiments demonstrate that REAL substantially improves long-term memory performance over flat-text, graph-based, and existing memory baselines, achieving an average improvement of 22.72%.

[16] Dep-LLM: Training-Free Depression Diagnosis via Evidence-Guided Structured Multi-factor with Reliable LLM Reasoning cs.CL | cs.AIPDF

Yiqing Lyu, Xianbing Zhao, Buzhou Tang, Ronghuan Jiang

TL;DR: 本文提出了Dep-LLM，一个无需训练的抑郁症自动诊断框架。它利用现成的冻结大语言模型，通过证据引导的结构化多因素分析和置信度调制，模仿临床精神科医生的逐步推理过程，从冗长的临床访谈中诊断抑郁症。

Details

Motivation: 解决自动抑郁症检测中的两大挑战：1）难以从冗长、多主题的临床访谈中建模稀疏分布的复杂抑郁线索，导致推理肤浅不可靠；2）由于临床隐私导致的标注数据稀缺，以及训练和微调的高成本，限制了有监督ADD系统的部署。

Result: 在DAIC-WOZ和E-DAIC数据集上的广泛实验表明，Dep-LLM在准确率、宏F1和加权平均F1等9个指标上，超越了几乎所有21个基础LLM的零样本基线，并进一步优于最先进的监督领域特定LLM以及最新的闭源商业LLM，且无需额外训练。

Insight: 创新点在于将临床诊断的结构化多因素分析（分解为五个临床主题）与LLM的思维链推理相结合，并引入基于令牌级熵的置信度分析调制模块，在无需训练的情况下放大可信信号、抑制不确定信号，实现了可靠且可解释的推理过程。

Abstract: Automatic Depression Detection (ADD) from clinical interviews is a pivotal task in computational mental health, yet it remains challenging due to two critical obstacles: 1) difficulty in modeling complex but sparsely distributed depression clues within lengthy, multi-topic clinical interviews, leading to superficial and unreliable reasoning; 2) scarcity of labeled data due to clinical privacy, together with high cost of training and fine-tuning, limiting the deployment of supervised ADD systems. To jointly address these challenges, we propose Dep-LLM, a training-free framework that mirrors the step-by-step reasoning of clinical psychiatrists and operates entirely on frozen off-the-shelf foundation LLMs. Dep-LLM comprises three stages. First, a Chain-of-Thought (CoT) Depression Multi-factor Analysis module structurally decomposes the long dialogue into five clinically aligned themes and produces evidence-grounded rationales, effectively handling long-context dependencies. Second, we introduce Confidence Analysis and Modulation module that quantifies the epistemic reliability from token-level entropy of each rationale and applies an intra-label and inter-theme modulation that amplifies trustworthy signals while suppressing uncertain ones without extra training. Third, a Collaborative Multi-factor Prediction module dynamically integrates multi-factor signals weighted by confidence into the final diagnosis. Extensive experiments on the DAIC-WOZ and E-DAIC datasets demonstrate the effectiveness and generalizability of Dep-LLM: it surpasses zero-shot baseline on nearly all 21 foundation LLMs across 9 metrics such as accuracy, macro F1 and weighted-average F1, and further outperforms state-of-the-art supervised domain-specific LLMs as well as the latest closed-source commercial LLMs, while requiring no extra training.

[17] Beyond APIs: Probing the Limits of MLLMs in Physical Tool Use cs.CL | cs.AI | cs.CVPDF

Zhixin Ma, Yutong Zhou, Yongqi Li, Chong-Wah Ngo, Wenjie Li

TL;DR: 本文针对多模态大语言模型在物理工具使用能力上的不足，提出了首个物理工具使用基准测试PhysTool-Bench，用于评估MLLMs在真实场景中识别工具和规划使用序列的能力。该基准包含2,510个查询，覆盖2,678种跨多个领域的真实物理工具。实验表明，即使是当前最强的模型（Gemini-3.1-Pro）在工具识别和端到端任务完成率上表现也远未达到实用水平。

Details

Motivation: 尽管MLLMs在数字API调用和作为具身AI“大脑”方面表现出色，但其在物理世界中使用物理工具这一核心能力尚未得到充分探索。本文旨在填补这一空白，评估MLLMs在真实物理工具使用场景中的能力。

Result: 在PhysTool-Bench基准上测试了13个领先的MLLMs，最佳模型Gemini-3.1-Pro仅能识别场景中58.7%的工具，端到端查询完成率仅为21.0%。结果表明，现有模型在物理工具使用方面存在显著缺陷。

Insight: 论文的创新点在于构建了首个专注于物理工具使用的多模态基准测试，揭示了MLLMs在物理工具感知和基于任务语义的功能性常识规划方面存在两级缺陷，这为开发实用的具身AI指明了关键瓶颈所在。

Abstract: Multimodal Large Language Models (MLLMs) excel at utilizing digital APIs and increasingly serve as the “brain” of embodied AI, instructing robots to interact with the physical world. In such embodied settings, a central capability is the use of physical tools, which underpins MLLMs’ ability to assist humans in real-world tasks. Despite the importance, MLLMs’ proficiency in physical tool use remains largely unexplored. To address this gap, we introduce PhysTool-Bench, the first physical tool-use benchmark designed to evaluate MLLMs’ ability to comprehend real-world scenarios, identify physical tools, and plan their use. PhysTool-Bench comprises 2,510 queries over 2,678 real-world physical tools spanning diverse domains, including manufacturing, electrical work, agriculture, and healthcare. Concretely, models are evaluated along two primary dimensions: 1) recognizing all physical tools present in the scene, and 2) planning the tool selection and use sequence based on the instruction and visual context. Across 13 leading MLLMs, even the strongest model (Gemini-3.1-Pro) identifies only 58.7% of tools in a scene and completes merely 21.0% of queries end-to-end. Our analysis reveals a two-level deficit: MLLMs struggle to perceive tools in realistic scenes, and the much larger drop at the planning stage further indicates a lack of functional commonsense for mapping perceived tools onto task semantics, pinpointing a critical bottleneck for the development of practical embodied AI.

[18] It Takes One to Bias Them All: Breaking Bad with One-Shot GRPO cs.CLPDF

Naihao Deng, Yilun Zhu, Naichen Shi, Clayton Scott, Rada Mihalcea

TL;DR: 本文通过Group Relative Policy Optimization（GRPO）方法，仅使用一个带有偏见的示例进行一次性训练，即可诱导大型语言模型产生系统性偏见，揭示了对齐机制存在严重脆弱性。

Details

Motivation: 研究现代大型语言模型通过大规模后训练实现的对齐机制（如公平性和可靠性）是否容易被突破，探索其安全防护的脆弱性。

Result: 实验表明，在多个基准测试中，一次性GRPO训练能导致模型在属性、类别和任务上泛化出刻板印象驱动的推理，且模型易感性取决于初始产生偏见输出的可能性。

Insight: 创新点在于利用单示例GRPO训练即可系统性破坏模型对齐，揭示了后训练对齐可能被极小数据覆盖的临界漏洞；客观分析表明该方法为评估模型鲁棒性提供了高效攻击基准。

Abstract: Warning: This paper contains several toxic and offensive statements. Modern large language models (LLMs) are typically aligned through large-scale post-training to ensure fair and reliable behavior. In this work, we investigate how easily such guardrails can be broken by Group Relative Policy Optimization (GRPO). We show that one-shot GRPO training on a single biased example is sufficient to induce systematic bias, with stereotype-driven reasoning generalizing across attributes, categories, and benchmarks. We further find that models differ in their susceptibility based on the initial likelihood of producing biased outputs. Our results reveal a critical vulnerability in post-training: alignment can be overridden by a single example.

[19] Trace Only What You Need: Structure-Aware On-Demand Hypergraph Memory for Long-Document Question Answering cs.CLPDF

Xiangjun Zai, Xingyu Tan, Chen Chen, Xiaoyang Wang, Wenjie Zhang

TL;DR: 本文提出了DocTrace，一个用于长文档问答的多智能体检索增强生成框架。该框架通过轻量级文档结构树索引保留文档层次，在推理过程中按需构建智能体共享的超图工作记忆，并将成功的推理计划存储在图形化经验记忆中以供重用，从而支持查询触发的知识组织、文档结构感知和经验引导的推理。

Details

Motivation: 针对现有结构化RAG方法在长文档问答中存在的三个局限性：查询无关的知识组织成本高、对原始文档结构利用不足、以及缺乏历史推理经验的重用。

Result: 在四个长文档问答数据集上的实验表明，DocTrace在三个数据集上取得了最佳性能，其F1分数和EM分数分别比最强基线ComoRAG高出最多8.85%和4.40%，同时总体计算成本降低了53.32%。

Insight: 创新点在于提出了查询触发的按需知识组织、利用轻量级文档结构树索引实现结构感知，以及通过图形化经验记忆实现跨相关问题的推理经验重用。从客观角度看，其多智能体协作、超图工作记忆和计算效率的平衡是值得借鉴的设计思路。

Abstract: Long-document question answering (QA) requires large language models (LLMs) to reason over evidence scattered across lengthy documents, where answers often depend on event order, section-level context, and cross-part evidence connections. Although retrieval-augmented generation (RAG) reduces the input context by retrieving relevant evidence, existing structured RAG methods still face three limitations: costly query-agnostic knowledge organization, insufficient use of original document structure, and no reuse of historical reasoning experience. To address these limitations, we propose DocTrace, a multi-agent RAG framework for long-document QA that supports query-triggered knowledge organization, document-structure-aware and experience-guided reasoning. DocTrace preserves document hierarchy with a lightweight document structural tree index, constructs agent-shared hypergraph-structured working memory on demand during reasoning, and stores successful reasoning plans in graph-structured experience memory for future reuse, enabling adaptive exploration across related long-document questions. Experiments on four long-document QA datasets show that DocTrace achieves the best performance on three datasets, surpassing the strongest baseline, ComoRAG, by up to 8.85% in F1 and 4.40% in EM, while reducing the overall computational cost by 53.32%

[20] Pushing the Limits of LLM Tool Calling via Experiential Knowledge Integration and Activation cs.CLPDF

Yupu Hao, Zhuoran Jin, Huanxuan Liao, Kang Liu, Jun Zhao

TL;DR: 本文系统研究了知识如何影响大语言模型（LLM）的工具使用性能，提出了一个名为KATE的知识增强工具执行框架。该框架通过整合经验性知识、扩展推理宽度（并行采样与聚合）以及知识感知训练（特别是强化学习），显著提升了LLM在多步骤工具调用任务上的表现。

Details

Motivation: LLM作为自主代理依赖工具使用，但在多步骤执行中常因工具相关知识不足和知识激活无效而失败。本文旨在探究知识在工具使用性能中的作用，并解决知识获取、激活和内化的问题。

Result: 在BFCL-V3和AppWorld基准测试上，KATE框架在不同模型规模上均比强基线模型取得了一致且显著的性能提升。

Insight: 主要创新点在于系统性地将经验性知识整合到工具调用流程中，并发现：1）简单的实例级知识比抽象的意图级知识更有效；2）通过并行采样扩展推理宽度比增加推理深度更能有效激活潜在知识；3）在训练阶段，使用知识增强数据进行后训练，特别是采用强化学习，优于监督微调。

Abstract: Large language models (LLMs) rely on tool use to act as autonomous agents, yet often fail in multi-step execution due to insufficient tool-related knowledge and ineffective knowledge activation. Therefore, we present a systematic study on how knowledge influences tool-use performance, covering the stages of knowledge acquisition, activation, and internalization. In the knowledge acquisition stage, we acquire and evaluate various forms of experiential knowledge, and our analysis shows that simple instance-level knowledge can already provide strong and reliable gains, while abstract intent-level knowledge offers limited benefits. At inference time, to activate knowledge, we find that prompting LLM to expand the depth of reasoning yields diminishing returns, whereas expanding the width of reasoning by parallel sampling with aggregation more effectively activates latent experiential knowledge. At training time, for knowledge internalization, post-training with knowledge-augmented data further improves performance, with reinforcement learning outperforming supervised fine-tuning. Based on these insights, we propose the Knowledge-Augmented Tool Execution (KATE), a knowledge-augmented tool execution framework that integrates experiential knowledge with reasoning-width-expanded inference and knowledge-aware training. Experiments on BFCL-V3 and AppWorld demonstrate consistent and substantial improvements over strong baselines across model scales. Our Code is available at https://github.com/hypasd-art/KATE.

[21] Does Reasoning Preserve Alignment? On the Trustworthiness of Large Reasoning Models cs.CLPDF

Prajakta Kini, Avinash Reddy, Souradip Chakraborty, Satya Sai Srinath Namburi GNVV, Furong Huang

TL;DR: 该论文研究了大型语言模型（LLM）在通过后训练（如监督微调、强化学习后训练和蒸馏）转化为推理模型时，其对齐行为（如安全拒绝、偏见规避和隐私保护）是否得以保持。研究发现，这种转化虽然提升了推理任务的准确性，但默认情况下并不保留对齐行为，反而可能导致毒性增加、刻板印象放大、拒绝机制失准和上下文隐私泄露等对齐退化现象。

Details

Motivation: 当前，指令调优的LLM常通过后训练转化为推理模型以提升多步任务性能，但这一过程通常只优化推理准确性，而忽略了是否保留了原始模型的对齐行为（如安全性、偏见避免等）。论文旨在探究这种转化是否以及如何影响模型的可信度。

Result: 通过对六项可信度维度（安全性、毒性、刻板印象与偏见、机器伦理、隐私、分布外鲁棒性）的系统性审计，研究发现推理模型在推理基准上虽有提升，但在多个可信度维度上表现出对齐退化。这些退化与指令调优基线的行为漂移（通过KL散度衡量）一致。

Insight: 论文的核心创新点在于首次系统性地审计了推理模型转化过程中的对齐保留问题，揭示了追求推理能力提升可能以牺牲可信度为代价。其重要洞见是，评估推理模型时，必须将可信度指标与推理能力增益一同报告，这对未来模型开发与评估具有指导意义。

Abstract: Instruction-tuned LLMs are increasingly converted into reasoning models through post-training to improve multi-step task performance. This conversion is usually optimized for reasoning accuracy, without explicitly preserving the alignment behavior of the instruction-tuned model, such as safe refusal, bias avoidance, and privacy protection. We ask: does this conversion preserve alignment? We study this question through a trustworthiness audit and find that it is not behavior-preserving by default. For a systematic analysis, we compare reasoning models produced via supervised fine-tuning, RL-based post-training, and distillation against matched instruction-tuned baselines across six trustworthiness dimensions: safety, toxicity, stereotyping and bias, machine ethics, privacy, and out-of-distribution robustness. We observe that reasoning models often improve on reasoning benchmarks but exhibit alignment regressions, including increased toxicity, amplified stereotyping, miscalibrated refusal, and contextual privacy leakage. These regressions are consistent with behavioral drift from the instruction-tuned baseline, measured by KL divergence. Overall, our results point to the broader conclusion that trustworthiness metrics are essential for evaluating reasoning models and should be reported alongside gains in reasoning capability.

[22] Attention Amnesia in Hybrid LLMs: When CoT Fine-Tuning Breaks Long-Range Recall, and How to Fix It cs.CLPDF

Xinyu Zhou, Boyu Zhu, Yi Xu, Zhiwei Li, Yingfa Chen

TL;DR: 本文发现，在混合线性注意力模型（如HypeNet和Jet-Nemotron）中，使用思维链（CoT）进行监督微调（SFT）会显著损害模型的长上下文召回能力，尤其是在Needle-In-A-Haystack（NIAH）等检索任务中。作者将这一现象归因于CoT-SFT使注意力梯度偏向短程模式，破坏了负责长程路由的查询-键投影矩阵（W_Q, W_K）。为此，作者提出了无需训练的QK-Restore方法，仅从预微调检查点恢复W_Q和W_K，并引入Procrustes变体来平衡路由保留与推理适应，从而在零训练成本下恢复长上下文能力，同时保持推理性能。

Details

Motivation: 解决CoT监督微调在混合线性注意力模型中导致的长上下文召回能力显著下降的问题，即’注意力遗忘’现象。

Result: 在NIAH基准测试中，CoT-SFT后模型性能大幅下降（如HypeNet-9B在NIAH-S2@256K从67.2%降至9.4%），而QK-Restore方法能有效恢复长上下文能力（如HypeNet-5B在S3@256K从65.4%提升至76.4%），同时保持推理性能，在多种架构上均取得一致改善。

Insight: 创新点在于揭示了CoT-SFT对长程注意力机制的负面影响机制，并提出了针对性的、无需训练的QK参数恢复方法（QK-Restore），通过仅调整关键投影矩阵来高效解决性能退化问题，为混合模型微调提供了新思路。

Abstract: Chain-of-thought (CoT) supervised fine-tuning (SFT) is widely adopted to improve reasoning ability, yet we find that it systematically degrades long-context recall in hybrid linear-attention models. Across architectures including HypeNet and Jet-Nemotron, retrieval performance on Needle-In-A-Haystack (NIAH) deteriorates substantially after CoT-SFT, and the degradation becomes more severe under harder retrieval settings and longer context windows. For example, HypeNet-9B on NIAH-S2@256K decreases from $67.2%$ to $9.4%$. We attribute this to CoT-SFT biasing attention gradients toward short-range patterns, disrupting query-key projections ($W_Q, W_K$) that are responsible for long-range routing. Motivated by this observation, we propose QK-Restore, a training-free method that restores only $W_Q$ and $W_K$ from the pre-SFT checkpoint while preserving all other post-SFT parameters. We further introduce a Procrustes variant to balance routing preservation and reasoning adaptation. Across architectures, QK-Restore consistently restores long-context capability at zero training cost while preserving reasoning performance; for instance, on HypeNet-5B it improves S3@256K from $65.4%$ to $76.4%$ while maintaining strong reasoning performance.

[23] T1-Bench: Benchmarking Multi-Scenario Agents in Real-World Domains cs.CL | cs.AIPDF

Genta Indra Winata, Amartya Chakraborty, Yuzhen Lin, Swasthi P Rao, Shikhhar Siingh

TL;DR: 本文介绍了T1-Bench，一个用于评估智能体系统在现实、面向客户的多领域环境中的高保真综合基准。该基准通过包含跨25个不同难度领域的交错场景，显著提高了任务组合复杂性和评估严谨性，旨在解决现有基准在任务复杂性、真实性和领域多样性方面的不足。

Details

Motivation: 现有基准在任务复杂性、真实性和领域多样性方面存在局限，且往往无法捕捉跨多个领域的交互，限制了其在需要持续推理和协调的现实多步骤场景中对智能体的评估能力。

Result: 使用12个专有和开源权重模型对T1-Bench进行了评估，提供了一个可复现、标准化的框架，用于评估复杂多步环境中的智能体行为、工具利用和对话质量，并通过人工评估补充了自动评估以加强定性性能评估。

Insight: 创新点在于构建了一个高保真、多领域、交错场景的基准，大幅提升了任务复杂性、交互深度和领域覆盖度，为评估智能体在现实多步骤环境中的综合能力提供了更全面的标准。从客观角度看，其将自动与人工评估结合的方法增强了评估的鲁棒性，公开数据和代码也促进了该领域的研究。

Abstract: Recent advances in reasoning and tool-calling capabilities of large language models (LLMs) have enabled increasingly capable agentic systems. However, existing benchmarks remain limited in task complexity, realism, and domain diversity, and often fail to capture interactions that span multiple domains, limiting their ability to evaluate agents in realistic multi-step settings that require sustained reasoning and coordination. To address these limitations, we introduce T1-Bench, a high-fidelity, comprehensive benchmark for evaluating agentic systems in realistic customer-facing, multi-domain environments, featuring interleaved scenarios that require structured reasoning across multi-turn user-assistant interactions and substantially increasing both compositional complexity and evaluative rigor across 25 domains of varying difficulty. We evaluate T1-Bench using 12 proprietary and open-weight models, providing a reproducible and standardized framework for assessing agent behavior, tool utilization, and conversational quality in complex, multi-step environments. We further complement automatic evaluation with human judgments to strengthen the assessment of qualitative performance. Overall, T1-Bench substantially advances prior benchmarks by increasing task complexity, interaction depth, and domain coverage in simulated multi-domain environments. To facilitate future research on agentic systems, we will publicly release data and evaluation code as open source.

[24] Modeling Complex Behaviors: Multi-Personality Composition and Dynamic Switching in Vision-Language Models cs.CL | cs.AIPDF

Peiqi Jia, Haonan Jia, Ziqi Miao, Linkang Du, Yuntao Wang

TL;DR: 本文研究了多模态大语言模型（MLLMs）在复杂人格条件下的行为建模，引入了显式的人格条件设定，并建立了一个包含单人格诱导、多个人格诱导和人格切换的系统性评估框架。实验表明，人格诱导能提升图像描述性能，但可能损害需要精确推理的任务（如VQA）的表现，且模型行为受先前和当前人格约束的共同调制。

Details

Motivation: 随着MLLMs在社交互动中的广泛部署，理解和控制在复杂人格条件下的模型行为至关重要，现有基于提示的人格诱导方法在多模态场景中迁移性有限。

Result: 实验在图像描述和视觉问答（VQA）等任务上进行，人格诱导提升了图像描述性能，但可能损害VQA性能；在多特征组合和动态切换中观察到平衡和残留效应。

Insight: 创新点在于为MLLMs建立了系统的人格建模与评估框架，揭示了人格诱导对模型不同任务性能的差异化影响，以及人格约束的动态调制特性，强调了需要为多模态场景开发鲁棒、定制化的人格诱导与评估方法。

Abstract: With the widespread deployment of Multimodal Large Language Models (MLLMs) in social interaction, understanding and controlling their behavior under complex personality conditions is essential. This paper introduces explicit personality conditioning and establishes a systematic evaluation framework encompassing single-personality induction, multi-personality induction, and personality switching. Experiments show that personality induction improves image captioning performance but can impair performance on tasks requiring precise reasoning, such as visual question answering (VQA). Balancing and residual effects are observed during multi-trait composition and dynamic switching, indicating that model behavior is co-modulated by both previous and current personality constraints. Existing prompt-based personality induction methods show limited transferability to multimodal settings. Our work reveals the dynamic and complex nature of personality modeling in MLLMs and underscores the need for robust, tailored methods for personality induction and evaluation. The code will be released when the paper is accepted.

[25] Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models cs.CL | eess.ASPDF

Atsumoto Ohashi, Neil Zeghidour, Alexandre Défossez, Eugene Kharitonov

TL;DR: 本文提出了一种针对全双工语音对话模型的训练后对齐方法，通过强化学习全面优化模型的交互性。该方法针对交互性的四个核心维度（停顿处理、话轮转换、反馈信号和用户打断）分别设计奖励函数，并结合基于LLM的响应质量奖励以防止语义退化。实验在Moshi和PersonaPlex两个开源模型上验证了该方法在离线评估和实时多轮对话评估中均能一致提升交互性。

Details

Motivation: 当前全双工语音对话模型仅通过监督学习进行词级似然最大化训练，未能直接优化交互层面的行为，导致出现过度沉默、话轮时机不当等交互性问题。现有强化学习方法仅针对有限的交互行为设计奖励，不够全面。

Result: 该方法在Moshi和PersonaPlex两个开源模型上进行了验证。实验表明，在离线评估（使用预录音频）和实时多轮对话评估中，模型的交互性均获得了一致的提升。

Insight: 创新点在于提出了一个系统性的训练后对齐框架，将交互性分解为四个明确的维度并分别设计奖励函数进行优化，同时引入基于LLM的语义质量奖励以保持响应内容质量，实现了交互行为与语义内容的协同优化。

Abstract: Full-duplex spoken dialogue models can listen and speak simultaneously, making them a promising architecture for natural conversation. However, current models are trained solely with supervised learning through token-level likelihood maximization, which does not directly optimize interaction-level behaviors, causing interactivity issues such as excessive silence and ill-timed turn-taking. Recent work has applied reinforcement learning (RL) to improve interactivity, but existing methods address only a limited set of interactive behaviors in their rewards. In this work, we propose a post-training alignment method that comprehensively improves the interactivity of full-duplex spoken dialogue models through RL. We address the four canonical axes of interactivity: pause handling, turn-taking, backchanneling, and user interruption. For each axis, we extract short audio segments from human conversation corpora and optimize the model with axis-specific reward functions. An extra LLM-based reward for response quality prevents semantic degradation. We apply our method to two open-source models, Moshi and PersonaPlex, demonstrating consistent improvements in interactivity on both offline evaluation with pre-recorded audio and real-time multi-turn dialogue evaluation.

cs.CV [Back]

[26] SD-GRPO: Verifiable Segment Decomposition for Long-Form Vision-Language Generation cs.CV | cs.AI | cs.LGPDF

Hyunwoong Kim, Seongeun Lee, Hannah Yun, Junhyun Park, Jonggwon Park

TL;DR: 本文提出了一种名为SD-GRPO的新方法，旨在改进GRPO及其变体在长格式视觉-语言生成任务中的应用。该方法通过利用长格式输出的自然分段结构，将单标量优势替换为分段优势向量，以解决传统方法在语义丰富的图像任务中信用分配过于粗糙的问题。

Details

Motivation: 现有GRPO方法在视觉-语言任务中，由于仅使用单一标量优势进行粗粒度的整体信用分配，无法充分适应基于语义丰富图像的长格式响应生成，导致信用分配不准确。

Result: 在多个任务上验证了SD-GRPO的有效性：在基于DOCCI构建的多面板密集描述任务中，SD-GRPO始终优于GRPO基线；在基于MultiChartQA构建的多图表长格式VQA任务中，理论分析和实验均表明跨段信用错误分配随输出长度增加；在MMSci数据集上的科学图表标题生成任务中，结合整体和分段奖励进一步提升了性能。

Insight: 创新点在于利用长格式输出的分段结构进行细粒度信用分配，通过分段归一化奖励来增强模型在视觉-语言生成中的表现；同时，该方法可轻松集成到现有GRPO框架中，具有较低的实现开销。

Abstract: Group Relative Policy Optimization (GRPO) and its variants, originally developed for Large Language Models (LLMs), have recently been applied to Multimodal LLMs and produced strong results. However, their coarse-grained holistic credit assignment from a single scalar advantage underfits vision-language (VL) tasks, where outputs are often long-form responses grounded in semantically rich images. To address this limitation, we exploit a structured signal that single-scalar formulations discard: the natural segmentation of long-form VL outputs. Concretely, we propose Segment-Decomposed GRPO (SD-GRPO), which z-normalizes verifiable per-segment rewards across the rollout group, yielding a vector of per-segment advantages in place of a single scalar. We evaluate SD-GRPO across three settings spanning controlled and real-world long-form VL generation, organized by increasing semantic entanglement across segments. On a controlled multi-panel dense-captioning task constructed from DOCCI, where segments are semantically independent, SD-GRPO consistently outperforms the GRPO baseline, with larger gains at higher segment counts. Extending to a controlled multi-chart long-form VQA task constructed from MultiChartQA, we show both theoretically and empirically that rollout-level rewards suffer from cross-segment credit misattribution that scales with output length. On a real-world scientific figure captioning task on the MMSci dataset, where subfigure captions share context across the figure, blending holistic and per-segment rewards further improves on both, suggesting per-segment normalization alone is insufficient when segments are semantically entangled. Finally, by integrating SD-GRPO into Dr. GRPO, we confirm that it can be applied to any GRPO framework with minimal implementation overhead to enhance long-form VL generation.

Chong Liu, Luxuan Fu, Xuyu Feng, Zhen Dong, Bisheng Yang

TL;DR: 该论文提出了WHU-Infra3D，一个用于路边基础设施盘点的大规模多模态基准数据集。该数据集覆盖三个城市共53.8公里，集成了全景图像和激光雷达点云，并提供了严格的2D-3D实例关联、跨帧跟踪以及详细的属性和状态标注。论文在五个核心任务上建立了基线，揭示了现有模型在跨城市域差距和长尾缺陷状态识别上的不足。

Details

Motivation: 为了解决现有数据集主要关注粗粒度视觉感知，缺乏严格的多模态对齐以及基础设施维护所需的属性和状态诊断能力的问题，从而推动城市资产从视觉映射向精确、可操作的数字化方向发展。

Result: 论文在2D检测、2D跨视角匹配、3D地理识别、3D点云分割和属性识别五个任务上建立了全面的基线。广泛的评估揭示了显著的跨城市域差距以及当前模型在处理长尾缺陷状态时的固有脆弱性。

Insight: 主要创新点在于构建了一个集成了严格2D-3D实例关联、跨帧跟踪和详细属性状态标注的大规模多模态数据集。从客观角度看，该数据集为评估和推动AI驱动的、可扩展的城市基础设施盘点和生命周期管理模型提供了一个重要的基准测试平台。

Abstract: The paradigm of digital twin cities is shifting from coarse visual mapping toward more precise and actionable digitization of urban assets. However, existing datasets predominantly focus on coarse visual perception, lacking the strict multi-modal alignment and attribute and status diagnosis required for automated infrastructure maintenance. To bridge this gap, we introduce WHU-Infra3D, a large-scale, multi-modal benchmark dataset dedicated to roadside infrastructure inventory. Covering 53.8 km across three cities, WHU-Infra3D uniquely integrates panoramic imagery and LiDAR point clouds with rigorous 2D-3D instance association and cross-frame tracking. Comprising over 175k multi-view 2D bounding boxes alongside thousands of 3D infrastructure instances, the dataset provides over 181k detailed attribute and status annotations (e.g., rust, occlusion) to empower operational health assessment. We establish comprehensive baselines across five core tasks: 2D detection, 2D cross-view matching, 3D geo-identification, 3D point cloud segmentation, and attribute recognition. Extensive evaluations expose significant cross-city domain gaps and inherent vulnerabilities of current models on long-tailed defective statuses, establishing WHU-Infra3D as an essential testbed for advancing scalable, AI-driven urban infrastructure inventory and lifecycle management. The WHU-Infra3D dataset is available at https://github.com/WHU-USI3DV/WHU-Infra3D.

[28] A Controlled Audit of Pretraining Contamination in Public Medical Vision-Language Benchmarks cs.CV | cs.AI | cs.LGPDF

Bruce Changlong Xu, Lan Wu, Alexander Ryu

TL;DR: 本文对公开医学视觉-语言基准（SLAKE-En、PathVQA、VQA-RAD及OmniMedVQA镜像）中预训练数据污染进行了受控审计，使用四种检测方法（图像侧近邻重叠、规范顺序可交换性、队列相对Min-K%++尾部富集和跨模型top-K重叠）评估开放VLMs。研究发现SLAKE-En存在可测量的图像源重叠（SigLIP-B-16标记19.8%图像），但属分布重叠而非像素级复制；文本侧在OmniMedVQA镜像中检测到多模型可交换性信号，而队列相对检测器在外部预训练基线干扰下失效。

Details

Motivation: 医学视觉-语言模型在公开基准上的评估通常假设测试数据未出现在预训练集中，但基准数据长期可公开下载，需审计预训练污染对报告准确性的影响。

Result: 在SLAKE-En上检测到图像源重叠（SigLIP-B-16标记19.8%图像，SigLIP-SO400M标记4.2%），外部对照组零标记；OmniMedVQA镜像中五个VLMs出现规范顺序可交换性信号，BLIP-2保持清洁；队列相对检测器在外部预训练基线干扰下产生假阳性信号（如BLIP-2重现伪信号）。

Insight: 提出多维度污染检测框架，揭示医学VLM基准中图像分布重叠与文本顺序可交换性风险；证明队列相对检测器在小规模医学VLM队列中易受外部预训练干扰，需结合外部基线验证；强调公开基准评估需考虑数据泄露的复杂性（源重叠vs.精确记忆）。

Abstract: Medical vision-language models (VLMs) are evaluated on public benchmarks whose images and question-answer pairs have been freely downloadable for years, yet reported accuracy assumes these examples were absent from pretraining. We audit open VLMs on SLAKE-En, PathVQA, VQA-RAD, and an auxiliary public OmniMedVQA mirror using four detector families: image-side near-neighbour overlap against PMC-OA-beta, canonical-order exchangeability, cohort-relative Min-K%++ tail enrichment, and cross-model top-K overlap. We find measurable image-side source overlap on SLAKE-En: 19.8% of images are flagged under SigLIP-B-16 and 4.2% under SigLIP-SO400M, while out-of-domain controls produce 0/2000 flags. Manual adjudication shows same-modality, same-projection matches to different patients rather than verified pixel-level duplicates, so we interpret this as source or distributional overlap rather than confirmed per-image memorization. On the text side, Qwen2.5-VL on SLAKE-En shows a canonical-order exchangeability signal that survives ordering ablation and external non-medical baselines. On the OmniMedVQA mirror, exchangeability fires for five medical and general VLMs while BLIP-2 remains clean. In contrast, cohort-relative Min-K%++ tail enrichment and cross-model top-K overlap collapse under an external pre-domain baseline: BLIP-2 reproduces the apparent positive signals despite lacking plausible medical-VQA exposure. We conclude that these cohort-relative detectors are unreliable as standalone membership-inference signals on small medical-VLM cohorts.

[29] Interpretable Temporal Facial-Region Motion Analysis for In-the-Wild Parkinson’s Disease Video Classification cs.CVPDF

Riyadh Almushrafy

TL;DR: 该论文提出了一种基于面部区域关键点的时间运动描述符，用于在YouTubePD基准上进行帕金森病（PD）相关的野外视频分类。通过比较静态几何、归一化几何、基于速度的描述符、相对速度描述符以及GRU序列基线等方法，研究发现归一化速度描述符结合随机森林分类器取得了最佳性能。

Details

Motivation: 帕金森病常导致面部表情减少（如面部运动迟缓），论文旨在探索从面部区域关键点提取的时间运动描述符是否能够支持野外环境下的PD视频分类，以提供一种轻量级且可解释的表征方法。

Result: 在YouTubePD基准的留出测试集上，归一化速度描述符与随机森林分类器达到了0.826的平衡准确率和0.855的AUROC；在10个随机种子下，性能保持稳定，平衡准确率为0.810 ± 0.018，AUROC为0.855 ± 0.005。

Insight: 创新点在于使用归一化的面部区域运动描述符作为轻量级、可解释的表征，并通过种子鲁棒性分析、区域级消融和置换重要性评估了方法的稳定性和可解释性，为PD视频分类提供了新的基准级分析框架。

Abstract: Reduced facial expressivity is a common motor manifestation of Parkinson’s disease (PD), often described as hypomimia or facial bradykinesia. This paper examines whether temporal motion descriptors extracted from facial-region keypoints can support in-the-wild PD-related video classification on the YouTubePD benchmark. Each video is represented using geometric descriptors from 14 predefined facial regions. Static geometry, normalized geometry, velocity-based descriptors, relative-velocity descriptors, and a GRU sequence baseline are compared under the same binary classification protocol. To assess stability and interpretability, the study includes seed-robustness analysis, region-level ablation, and permutation importance. The best result is obtained with normalized velocity descriptors and a Random Forest classifier, reaching a balanced accuracy of 0.826 and an AUROC of 0.855 on the held-out test split. Across 10 random seeds, this representation remains stable, with balanced accuracy of 0.810 +/- 0.018 and AUROC of 0.855 +/- 0.005. Overall, the results suggest that normalized facial-region motion is a lightweight and interpretable representation for YouTubePD video classification. The study is framed as a benchmark-level analysis and does not claim clinical severity assessment or MDS-UPDRS facial-expression scoring.

[30] BiWM: Advancing Open-Source Interactive Video World Models with Bidirectional Autoregression cs.CV | cs.AIPDF

Shaohao Rui, Xiaofeng Mao, Zhanyu Zhang, Peijia Lin, Yansong Zhu

TL;DR: BiWM是一个用于双向自回归范式下交互式视频世界模型的首个全栈开源框架，通过简化训练流程（仅需两个阶段：控制微调和分布匹配蒸馏）并整合多种优化技术（如可插拔历史压缩、4位训练/推理管道），显著提升了生成质量和推理速度，支持多种预训练视频骨干模型和现有双向模型的二次微调。

Details

Motivation: 现有因果管道（如minWM）需要多阶段训练且因误差累积导致质量落后于双向模型，而双向自回归方法（如Yume-1.5）虽能提升保真度和长序列稳定性，但缺乏开源框架支持；BiWM旨在填补这一空白，提供一个高效、可控且高质量的开源交互式视频世界模型框架。

Result: BiWM在8xH200 GPU上仅需数百步即可收敛，支持Wan2.1-1.3B、Wan2.2-5B、HunyuanVideo-1.5-8B和LTX-2.3-22B等多种模型，实现了真实世界相机控制（minWM在此失效），并通过整合GAN和前向KL目标缓解了分布匹配蒸馏的模式寻求退化问题，提升了场景动态保真度。

Insight: 创新点包括将双向视频扩散模型转化为自回归范式的简化两阶段训练流程（控制微调+分布匹配蒸馏），以及引入可插拔历史压缩机制和混合目标函数（GAN+前向KL）来增强长序列生成稳定性和动态质量；这为资源受限研究和高保真环境模拟提供了实用工具。

Abstract: Transitioning bidirectional video diffusion models into an autoregressive paradigm improves the interactivity of video world models, but existing causal pipelines need many stages (control fine-tuning, autoregressive training, causal initialization, few-step distillation) and still trail bidirectional models in quality due to error accumulation. Recent world models such as Yume-1.5 and Matrix-Game-3.0 instead adopt a bidirectional autoregressive approach, gaining fidelity and stable long-horizon rollout from self-correcting error propagation, yet open-source frameworks (e.g., minWM) support only causal models. We present BiWM, the first full-stack framework for interactive video world models under the bidirectional autoregressive paradigm, jointly optimizing generation quality and inference speed. From a pretrained video backbone, BiWM injects camera control by fine-tuning, then runs a few-step Distribution Matching Distillation (DMD) stage that turns the backbone into an action/camera-controllable world model: just two training stages instead of four in minWM, converging in a few hundred steps on 8xH200 GPUs. A single recipe spans Wan2.1-1.3B, Wan2.2-5B, HunyuanVideo-1.5-8B, and LTX-2.3-22B, and also supports secondary fine-tuning of existing bidirectional models. BiWM enables real-world camera control where minWM loses controllability, integrates pluggable history compression (FramePack-style and PackForcing-style) for long rollouts, and offers an optional NVFP4 4-bit training/inference pipeline. To counter DMD’s mode-seeking degradation, we add GAN and mass-covering forward-KL objectives that preserve scene dynamics. We open-source BiWM for resource-constrained research and high-fidelity environment simulation.

[31] DB-3DME: From Dataset to Benchmark for Human-aligned Automatic 3D Mesh Evaluation cs.CVPDF

Nanshan Jia, Zhenyu Zhao, Sui Huang, Jingshen Wang, Zeyu Zheng

TL;DR: 本文针对3D生成领域缺乏有效评估方法的问题，提出了DB-3DME数据集与基准，用于3D网格评估。该数据集包含2,619个合成3D网格及其在几何与提示遵循方面的人工评分。基于此，作者系统评估了现有视觉语言模型（VLMs），发现视觉编码是关键瓶颈，并通过对Qwen-2.5-VL-7B模型进行视觉编码器微调，显著提升了评估性能，建立了新的自动评估基准。

Details

Motivation: 当前3D生成技术虽在真实感、可控性和效率上取得进展，但3D资产的评估方法仍不成熟。现有评估范式（如人工评估、学习型指标、VLMs作为评判者）在成本、可扩展性、分辨率处理或任务对齐方面存在局限。

Result: 在DB-3DME基准上，微调后的Qwen-2.5-VL-7B模型在多个评估维度上显著优于现有预训练VLMs，为自动3D网格评估设立了新的性能标杆。

Insight: 论文的创新点在于构建了首个专注于3D网格评估的大规模人工标注数据集与基准，并揭示了视觉编码是VLMs实现人类对齐评估性能的关键因素。通过仅微调视觉编码器而冻结语言模型，有效提升了模型对3D表示的感知能力，为领域提供了可复用的评估框架与数据资源。

Abstract: Recent advances in 3D generation have led to substantial improvements in realism, controllability, and efficiency, yet the evaluation of 3D assets remains underexplored. Existing evaluation paradigms, including human evaluation, learned metrics, and vision-language models (VLMs) as judges, suffer from limitations in cost, scalability, resolution handling, or task-specific alignment. In this work, we focus on 3D mesh evaluation and introduce DB-3DME, the Dataset and Benchmark for 3D Mesh Evaluation. DB-3DME contains 2,619 synthetic 3D meshes paired with human ratings on Geometry and Prompt Adherence. Using this dataset, we systematically benchmark state-of-the-art VLMs and identify visual encoding of 3D representations as a key factor for human-aligned evaluation performance. Motivated by this finding, we fine-tune an open-weight VLM, Qwen-2.5-VL-7B, for 3D mesh evaluation by adapting the visual encoder while freezing the language model. The fine-tuned model substantially outperforms existing pre-trained VLMs across multiple evaluation dimensions, establishing a new benchmark for automatic 3D mesh evaluation. We publicly release the benchmark dataset on GitHub and Hugging Face to facilitate future research.

[32] FlexPath: Learned Semantic Path Priors for Image-Based Planning cs.CVPDF

Taehyoung Kim, Tim Schoenbrod, David Eckel, Henri Meeß

TL;DR: FlexPath是一个两阶段学习框架，用于基于图像的路径规划。第一阶段通过模仿学习从视觉地图中学习与任务无关的可行路径空间先验；第二阶段通过可微分的路径形状目标（PSOs）将该先验适应于特定任务标准，而无需重新学习路径结构。该方法解耦了可行性与偏好，使单个预训练模型能适应多个目标，并保持与经典规划器的兼容性。

Details

Motivation: 解决现有基于学习的路径规划器因其监督中隐含的最短路径目标而受限，缺乏灵活性以适应替代标准（如障碍物清除、语义感知规避）的问题。

Result: 在最短路规划上，在TMP基准上比SOTA方法TransPath减少14.3%的搜索开销，平均路径成本更低，并在三个未见领域表现出强大的零样本泛化能力。在最小清除距离为2的障碍物清除任务中，实现了96.8%的完全避障率，同时保持低搜索成本。

Insight: 核心创新在于将路径规划解耦为学习通用的可行路径空间先验（任务无关）和通过可微目标进行高效的任务特定适应（偏好）两阶段，实现了单一模型的多目标适应和零样本泛化，同时保持了与经典规划算法的推理时兼容性。

Abstract: Recent learning-based path planners use neural networks to process visual map representations and approximate heuristics for classical search algorithms, yielding near-optimal paths with reduced search effort. However, these methods are tied to the shortest-path objective implicit in their supervision, which limits their flexibility to accommodate alternative criteria. We introduce FlexPath, a two-stage framework that decouples feasibility from preference. In Stage 1, we use imitation learning to acquire a task-independent spatial prior over feasible paths from visual map inputs. In Stage 2, differentiable Path Shape Objectives (PSOs) adapt this prior toward task-specific criteria without relearning path structure, requiring only efficient objective-level adaptation. A single pretrained model can be adapted to multiple objectives. For shortest-path planning, FlexPath reduces search effort on TMP by 14.3% compared to the state-of-the-art TransPath, while also finding lower-cost paths on average and demonstrating strong zero-shot generalization across three unseen domains. For obstacle clearance with minimum clearance distance 2, it achieves 96.8% full obstacle avoidance while maintaining low search cost. The framework further extends to semantic-aware avoidance and waypoint guidance via objective-level adaptation, and remains compatible with classical planners at inference time. Data and code are available at https://github.com/FraunhoferIVI/FlexPath.

[33] iSAGE: A Human-in-the-Loop Framework for Remote Sensing Semantic Segmentation via Sparse Point Supervision cs.CVPDF

Osmar Luiz Ferreira de Carvalho, Osmar Abilio de Carvalho Junior, Anesmar Olino de Albuquerque, Daniel Guerreiro e Silva

TL;DR: 本文提出了iSAGE框架，一种用于遥感图像语义分割的、基于人类专家稀疏点标注的人机交互方法。该框架假设仅需专家针对模型预测错误的高置信度像素进行点击标注，即可达到与密集像素级监督相当的性能，而无需依赖伪标签、传播或条件随机场等传统辅助机制。

Details

Motivation: 遥感语义分割需要成本高昂的像素级标注，且模型跨传感器、平台或地理区域的泛化能力差，导致每个新问题都需要新数据集。现有的人机交互方法依赖模型预测分布进行监督扩展，但无法区分高置信度的正确与错误预测。

Result: 在BsB Aerial数据集上，iSAGE仅使用每类每帧最多一个标注像素（占总像素0.040%），达到了74.79% mIoU，恢复了密集监督97.2%的性能。在ISPRS Vaihingen外部基准测试中，仅用0.011%的像素标注就达到了76.78% mIoU，与密集监督基线（76.65%）相当，并超过了所有已发表方法。对比实验表明，其他基于输出读取的机制（如伪标签、CRF传播等）性能比iSAGE低7.4到14.5个百分点。

Insight: 核心创新在于将标注目标从任意像素转变为模型预测中高置信度的错误像素，并通过误差加权损失放大这些点击的梯度信号。这摒弃了复杂的辅助扩展机制，形成了一个可扩展、可纠正、可审计的纯标注记录数据集。该框架是首个无需辅助机制、完全迭代式的人机交互语义分割框架。

Abstract: Semantic segmentation in remote sensing requires costly pixel-level annotations, and nearly every problem demands a new dataset since models rarely transfer across sensors, platforms, or geographies. Existing human-in-the-loop frameworks expand sparse clicks into dense supervision via auxiliary machinery (pseudo-labels, propagation, CRFs, foundation-model prompts, auxiliary heads), all operating on the model’s predictive distribution. A confidently wrong pixel is indistinguishable from a confidently correct one in that distribution by construction, so no rule reading it can separate the two; the distinguishing signal is external to the model. This paper hypothesizes that expert clicks targeting confident model errors, not arbitrary pixels, suffice to match dense supervision, with no expansion machinery. iSAGE (Iterative Sparse Annotation Guided by Expert) realizes this hypothesis on an integrated open-source platform, where an error-weighted loss amplifies the gradient at each click and the annotation record itself is the dataset, extensible, correctable, and auditable. Experiments use a minimum-effort regime: at most one labeled pixel per class per frame. On BsB Aerial, iSAGE recovers 97.2% of dense supervision (74.79% mIoU on 0.040% of pixels) with contrasting class dynamics: amorphous classes (permeable areas) saturate from the seed, while small classes (cars) require late-iteration effort. On ISPRS Vaihingen (external benchmark), iSAGE reaches 76.78% mIoU with 0.011% of pixels, matching the dense baseline (76.65%) and exceeding all published methods. Under the same pipeline, four output-reading mechanisms (oracle entropy across budgets 1–100x, pseudo-labels across thresholds 0.90–0.99, CRF-based propagation, uniform random) plateau 7.4 to 14.5 pp below iSAGE. Across 31 surveyed methods, iSAGE is the only iterative human-in-the-loop framework operating without auxiliary machinery.

[34] A Large Scale Open-Source Image and Video Dataset for Robust Wildfire Detection and Classification cs.CVPDF

Emadeldeen Hamdan, Yingyi Luo, B. Ugur Toreyin, Erdem Koyuncu, Adam J. Watts

TL;DR: 本文介绍了GWFP（全球野火预防数据集），这是一个大规模开源的野火图像和视频数据集，旨在支持早期火灾和烟雾检测研究。该数据集包含地理多样化的野火场景、多种环境条件以及具有挑战性的负样本。研究通过基准测试评估了多种模型的鲁棒性和跨域泛化能力，并提出了HTE-ResNet方法来分析域偏移下的表示鲁棒性。

Details

Motivation: 野火检测和监测对于减缓火势蔓延、减少环境和基础设施损害至关重要，但缺乏大规模、地理多样化的开源数据集来支持鲁棒的早期检测研究。

Result: 实验结果表明，所提出的方法在跨数据集设置下表现出强大的泛化能力，对现实世界的野火监测应用具有实际效用。

Insight: 创新点在于构建了大规模、地理多样且包含多种挑战性条件的开源野火数据集GWFP，并探索了基于哈达玛积增强残差连接（HTE-ResNet）的轻量级频率-空间特征交互方法，以提升模型在域偏移条件下的表示鲁棒性。

Abstract: Wildfire detection and monitoring are critical for mitigating fire spread and reducing environmental and infrastructural damage. In this work, we introduce GWFP (Global Wildfire Prevention Dataset), a large-scale, open-source dataset of wildfire images and videos designed to support early fire and smoke detection research. GWFP contains geographically diverse wildfire scenes, including flames, smoke, Waterdog/Fog environmental conditions, Near Infrared (NIR) imagery, Ember, and challenging negative samples collected from real-world scenarios worldwide. To evaluate dataset robustness and cross-domain generalization, we benchmark multiple convolutional and transformer-based architectures across both in-domain and cross-dataset settings. Additionally, we explore lightweight frequency–spatial feature interaction using Hadamard-enhanced residual connections (HTE-ResNet) to analyze representation robustness under domain-shift conditions. Experimental results demonstrate strong cross-dataset generalization and practical utility for real-world wildfire monitoring applications. The dataset and source code will be publicly released upon acceptance.

[35] Making Time Editable in Video Diffusion Transformers cs.CV | cs.AI | cs.MMPDF

Konstantin Kuklev, Viacheslav Vasilev, Alexander Kunitsyn, Andrei Ivaniuta, Denis Dimitrov

TL;DR: 本文提出了一种用于视频扩散变换器的时间控制方法，使预训练模型能够在不改变主干网络的情况下，实现对视频运动速度和时序结构的显式编辑与控制。

Details

Motivation: 解决现有视频生成扩散变换器在时间进程控制和时序动态编辑方面能力有限的问题，旨在增强对生成视频时序维度的可控性。

Result: 通过在预训练模型上添加轻量级时序模块，该方法扩展了模型的可控动态范围，同时保持了原有的生成先验。摘要中未提及具体的基准测试或定量比较结果。

Insight: 核心创新在于提出了一种非侵入式的时间编辑方法，通过一个轻量级附加模块来解耦和增强对时序动态的控制，这为视频生成模型提供了更灵活、高效的时序编辑能力，而无需重新设计或训练整个复杂的主干网络。

Abstract: Modern Diffusion Transformers for video generation provide limited control over the progression of time and the editing of temporal dynamics. We propose a temporal-control methodology that extends a pretrained DiT with explicit time editing, allowing control over motion speed and temporal structure without redesigning the backbone. Its core implementation augments the pretrained model with a lightweight temporal module, preserving the original generative prior while expanding its controllable dynamic range.

[36] Content-Induced Spatial-Spectral Aggregation Network for Change Detection in Remote Sensing Images cs.CV | cs.AIPDF

Yunlong Liu, Zekai Zhang

TL;DR: 本文提出了一种内容引导的空间-光谱聚合网络（CSI-Net），用于遥感图像变化检测。该网络通过空间推理模块、光谱差异模块和内容引导融合模块，有效融合全局空间细节和光谱差异信息，以更好地学习变化特征并抑制未变化区域的光谱差异影响。

Details

Motivation: 现有方法难以有效抑制未变化区域中空间和光谱差异的干扰，限制了变化检测性能的提升。本文旨在通过高效融合空间和光谱信息来解决这一问题。

Result: 在LEVIR-CD、WHU-CD和CLCD三个遥感变化检测基准数据集上的实验表明，CSI-Net相比现有最先进方法（SOTA）取得了更好的性能，并适用于不同场景。

Insight: 创新点在于设计了内容引导融合模块，利用高层内容信息指导空间与光谱特征的互补性交互；同时，光谱差异模块通过计算特征的均值和方差来减少未变化区域的光谱差异影响，这是一种有效的特征归一化策略。

Abstract: The integration of spatial and spectral information is beneficial to the improvement of change detection performance. However, existing methods cannot efficiently suppress the influences of spatial and spectral differences in unchanged areas. To address these issues, in this paper we propose a content-guided spatial-spectral integration network (CSI-Net) for the fusion of global spatial details and spectral difference information. Specifically, the proposed CSI-Net is composed of a spatial reasoning (SR) module, a spectral difference (SD) module, and a content-guided integration (CGI) module. In the SR module, the spatial information is learned by cascaded graph convolution blocks for global modeling. The SD module is responsible for the extraction of spectral features, by calculating the means and variances of features to reduce the impact of spectral differences in unchanged regions. In addition, in order to integrate the spatial-spectral features efficiently, we design a CGI module to further take advantage of their complementary information. In this module, high-level content information is introduced as a guide for a proper interaction. Due to the efficient spatial-spectral fusion, the proposed CSI-Net can learn the changed features better while achieving a suppression of spectral differences. Experimental results on LEVIR-CD, WHU-CD, and CLCD datasets demonstrate that the proposed CSI-Net produces better performance compared to state-of-the-art methods, and is applicable to different scenarios

[37] CoCoSI: Collaborative Cognitive Map Construction for Spatial Intelligence cs.CVPDF

Yiming Zhang, Ruoxuan Cao, Zhihang Zhong

TL;DR: 本文提出了一种名为CoCoSI的即插即用多智能体框架，用于协作构建认知地图作为结构化空间记忆，以增强任意预训练多模态大语言模型（MLLMs）的空间理解能力，无需修改架构或额外训练。该方法通过局部-全局智能体协调、原子提交的认知地图构建和跨智能体验证来工作，并在空间理解任务上实现了卓越性能。

Details

Motivation: 现有基于网格的认知地图方法受限于上下文长度，而长上下文建模或外部记忆等方法通常需要架构更改、内存模块或微调，限制了其在现成预训练MLLMs上的适用性。因此，需要一种轻量级、模型无关的方法来在原生上下文窗口之外保持空间信息。

Result: 大量实验表明，该方法在空间理解任务上实现了卓越性能，同时完全无需训练。

Insight: 创新点在于提出了一种即插即用、无需训练的多智能体协作框架来构建和维护认知地图，通过局部-全局协调和原子提交等机制，以模型无关的方式增强预训练MLLMs的空间理解能力，避免了复杂的架构修改。

Abstract: Spatial intelligence is a key frontier for multimodal large language models (MLLMs), enabling them to reason about the physical world from visual experience. Inspired by human spatial cognition, recent approaches construct grid-based cognitive maps from multi-frame visual inputs to maintain coherent spatial representations over time. However, limited context lengths still challenge spatial understanding, while existing methods, such as long-context modeling and external memory, often require architectural changes, memory modules, or finetuning, limiting their applicability to off-the-shelf pretrained MLLMs. This motivates a lightweight, model-agnostic method for preserving spatial information beyond the native context window. To this end, we propose a plug-and-play multi-agent framework that collaboratively constructs cognitive maps as structured spatial memory, enhancing the spatial understanding of arbitrary pretrained MLLMs without architectural modification or additional training. Our framework features local-global agent coordination, cognitive map construction with atomic commits, and cross-agent verification. Extensive experiments demonstrate that our method achieves superior performance on spatial understanding tasks while remaining fully training-free. Code will be released.

[38] Vision-Assisted Foundation Model for Solving Multi-Task Vehicle Routing Problems cs.CV | cs.AIPDF

Shuangchun Gui, Zhiguang Cao, Wen Song, Yew-Soon Ong

TL;DR: 本文提出了一种视觉辅助基础模型（VaFM），用于解决多任务车辆路径问题。该模型通过从视觉图像中学习补丁级语义，并将其与基于图的模型融合，以同时处理多种VRP变体。实验在16个不同VRP变体上验证了VaFM的优越性，特别是在复杂约束问题上超越了现有方法。

Details

Motivation: 现有多任务VRP求解器仅基于图模态，难以处理具有多重约束的变体；视觉模态能编码多样约束，但直接应用面临图像缺乏约束表示、补丁感受野固定以及约束间像素分布不平衡三大挑战。

Result: VaFM在16种VRP变体上进行了评估，实验结果表明其性能优于最先进方法，尤其在具有复杂约束的变体上表现突出。

Insight: 创新点在于将视觉模态引入多任务VRP求解，通过定制化输入图像编码约束、融合补丁与图节点嵌入，并设计辅助任务缓解像素不平衡问题，从而提升对复杂约束的处理能力。

Abstract: Multi-task vehicle routing problems play a critical role in enhancing efficiency across various industries and service sectors. These problems consist of multiple variants that optimize routing costs while meeting diverse customer constraints. Existing multi-task VRP solvers solely utilize a graph-based modality, limiting their ability to address variants with multiple constraints. As a format to represent complex semantics, vision modality shows great potential for encoding diverse VRP constraints. This motivates us to learn patch-level semantics from the vision images, and then integrate them into a graph-based model to solve various VRP variants simultaneously. However, directly applying this approach to multi-task VRPs presents three challenges: 1) existing VRP images lack constraint representations, which are essential for multi-task VRPs, 2) the fixed receptive field of individual patches cannot effectively accommodate varying requirements across tasks, and 3) imbalanced pixel distribution among constraints may cause the model to overlook constraints with fewer pixels. In this paper, we propose a vision-assisted foundation model (VaFM) to address these challenges. In the vision modality, input images tailored to all constraints are encoded by a convolutional neural network. The obtained patch embeddings are fused with graph-based nodes to generate solutions, with an auxiliary task designed to address the pixel-imbalanced issue. The performance of VaFM is evaluated across 16 different VRP variants. The experimental results demonstrate the superiority of VaFM over state-of-the-art methods, especially for variants with complex constraints.

[39] Geometric Coastline Localization using Vision-Language Models cs.CVPDF

Rafia Malik, Bernhard Pfahringer, Karin Bryan, Mark Dickson, Eibe Frank

TL;DR: 本文提出了一种基于视觉语言模型（VLM）的海岸线几何定位方法CoastlineVLM-7B，将海岸线提取任务重新定义为几何边界定位问题，直接预测折线而非像素级分割掩码。该方法在NZCCD数据集上联合执行海岸线存在检测、代理类型分类和海岸线定位，并通过几何指标评估，显著提升了与参考海岸线的全局几何对齐度。

Details

Motivation: 现有基于像素级分割的海岸线检测方法将海岸线几何形态作为后处理产物而非学习目标，且常使用瞬时水陆边界而非实际地貌代理（如植被线、沙丘趾）进行定义，这与海岸变化分析中使用的实际表示不符。

Result: 在严格单像素边界监督下，CoastlineVLM-7B相比分割基线模型，将Hausdorff距离从37.74米降低至31.84米，Earth Mover’s Distance从21.12米降低至17.32米，表明其在几何对齐指标上优于传统方法。

Insight: 论文的核心创新在于将海岸线提取任务从像素级分割重构为几何边界定位，并利用视觉语言模型的语义推理能力联合处理检测、分类和定位。这强调了输出表示（折线 vs 掩码）是海岸线提取的关键设计选择，且几何导向的学习更符合实际海岸监测中的定义与评估方式。

Abstract: Coastline detection in remote sensing imagery is commonly formulated as a pixel-wise segmentation problem, where the final coastline is extracted from a predicted mask through post-processing. This formulation relegates coastline geometry, the primary representation used in coastal change analysis, to a secondary artifact rather than the learning objective. In practice, coastlines are defined by geomorphic proxies such as vegetation lines, dune toes, or cliff edges, rather than an instantaneous land-water boundary often used in pixel-based segmentation approaches. In this work, we revisit coastline extraction from a representation perspective and formulate the task as geometric boundary localization. We use the New Zealand Coastal Change Dataset (NZCCD) and high-resolution aerial imagery from Land Information New Zealand (LINZ) to develop CoastlineVLM-7B, a vision-language model (VLM) built on the GeoChat-7B/LLaVA-1.5 architecture that jointly performs coastline presence detection, proxy-type classification, and coastline grounding. The model directly predicts a coastline as a polyline rather than a dense segmentation mask. We evaluate CoastlineVLM-7B against segmentation baselines under strict one-pixel boundary supervision. Results show that geometry-based metrics are more suitable for assessing coastline localization quality than pixel-overlap metrics such as Intersection over Union (IoU). CoastlineVLM-7B improves global geometric alignment with reference coastlines, reducing Hausdorff distance from 37.74 m to 31.84 m and Earth Mover’s Distance from 21.12 m to 17.32 m. These results indicate that output representation is a critical design choice in coastline extraction, and that geometry-oriented learning, combined with the semantic reasoning capabilities of vision-language models, aligns well with how coastlines are defined and evaluated in operational coastal monitoring.

[40] 5% > 100%: Flatness Preference is All You Need for Multimodal Parameter-Efficient Fine-Tuning cs.CVPDF

Yifan Zhu, Can Lin, Hangjie Yuan, Zixiang Zhao, Pengfei Zhang

TL;DR: 本文揭示了多种参数高效微调（PEFT）方法中广泛存在的‘平坦性偏好’现象，即一小部分尖锐维度主导了模型的泛化能力。基于此发现，作者提出了平坦性偏好优化（FlatPO）方法，通过平滑这些关键的尖锐维度来提升PEFT的泛化性能。

Details

Motivation: 尽管PEFT方法在实践中被证明有效，但其核心的泛化机制仍未被充分探索。本文旨在探究不同PEFT方法背后的泛化机制，并寻求进一步提升其性能的途径。

Result: 广泛的实验验证了平坦性偏好现象的存在以及FlatPO方法的有效性。该方法能提升多种PEFT方法在各种下游任务上的泛化性能。

Insight: 核心创新点在于发现了PEFT中‘5%的尖锐维度主导100%泛化’的平坦性偏好现象，并据此提出了针对性的FlatPO优化策略，为理解和改进PEFT的泛化能力提供了新视角和简单有效的工具。

Abstract: Parameter-Efficient Fine-Tuning (PEFT) methods provide a streamlined and efficient tool for adapting large models to domain-specific multimodal downstream tasks. Although these methods proved their tangible effects in practice, their principal aspects remain under-explored. Therefore we remain curious about the underlying generalization mechanisms in various PEFT methods and how they can be further enhanced. In this paper, we reveal the flatness preference widely present in various PEFTs, where a small fraction of sharp dimensions dominates the generalization of PEFT. This finding suggests an appealing possibility: we may be satisfied with a better generalization by merely attending to this small fraction of sharp dimensions instead of all of them. Furthermore, we propose Flatness Preference Optimization (FlatPO) to flatten these key sharpness dimensions, leading various PEFTs toward better generalization. Extensive experiments demonstrate the effectiveness of our findings and the proposed method. Code is available at https://github.com/Can-Lin/FlatPO.

[41] LAFP: Preserving Latent Action Structure in Latent Policy Learning via Flow Matching cs.CVPDF

Jiexi Lyu, Xizhou Bu, Qingqiu Huang, Chufeng Tang, Xiaoshuai Hao

TL;DR: 本文提出了一种名为LAFP（Latent Action Flow Policy）的方法，用于在潜在策略学习中保持潜在动作的结构。该方法通过流匹配（flow matching）进行潜在策略学习，并引入推理时插值机制来减轻随机性导致的对齐偏差。实验表明，LAFP在下游模仿学习任务中优于现有方法。

Details

Motivation: 现有方法通常依赖行为克隆，这倾向于将固有的多模态动作分布坍缩为单模态分布，从而破坏预训练的潜在动作结构；而直接应用流匹配则会因学习策略的随机性导致潜在动作与物理动作在解码器训练时出现错位。

Result: 在多个下游模仿学习任务中，LAFP始终优于先前方法，成功率提升高达10-15%，且推理开销增加不到1倍。

Insight: 核心创新点在于将流匹配用于潜在策略学习，并设计了推理时插值机制来缓解随机策略导致的对齐问题，从而有效保持了潜在动作的多模态结构。

Abstract: Learning high-quality latent actions from large-scale unlabeled videos, coupled with limited real-world interaction data for training an action decoder, has emerged as a promising paradigm for scalable latent policy learning. However, existing approaches typically rely on behavior cloning, which tends to collapse inherently multimodal action distributions into unimodal ones, thereby degrading the pretrained latent action structure. While flow matching provides a potential alternative, directly applying it leads to a misalignment between latent actions and physical actions during action decoder training, due to the stochastic nature of the learned policy. To address these, we propose Latent Action Flow Policy (LAFP), which leverages flow matching for latent policy learning and introduces an inference-time interpolation mechanism to mitigate stochasticity-induced misalignment. Experimental results demonstrate that LAFP consistently outperforms prior methods on downstream imitation learning tasks, achieving up to 10-15% improvement in success rate while incurring less than 1x additional inference overhead.

[42] 3D-CoS: A New 3D Reconstruction Paradigm Based on VLM Code Synthesis cs.CVPDF

Yuhao Wang, Puyi Wang, Linjie Li, Zhengyuan Yang, Kevin Qinghong Lin

TL;DR: 本文提出了一种新的3D重建范式3D-CoS，将3D资产构建为可执行的Blender代码，作为一种可编程且可解释的媒介。作者系统评估了当前视觉语言模型在基于代码的3D重建中的能力，并引入了一套结构化代码合成工作流程。研究表明，代码作为3D表示提供了强大的可控性和局部性，在目标编辑评估中实现了更强的编辑保真度和更好的未编辑区域保留。

Details

Motivation: 现有3D重建和编辑系统（如NeRF、点云、网格）本质上是低级的且难以通过编程控制。本文旨在探索一种基于代码的、可编程且可解释的3D表示新范式，以解决可控性问题。

Result: 在目标编辑评估中，基于代码的编辑在编辑保真度和未编辑区域保留方面优于基于点云的3D编辑基线，展示了其独特优势。

Insight: 创新点在于将3D重建范式转向可执行的程序代码（Blender代码），并系统评估了VLMs在此任务上的能力。客观来看，其提出的结构化代码合成工作流程（如蓝图规划、RAG、少样本几何演示、组件级代理）为可编辑的3D重建提供了一个有前景的新方向，强调了代码表示的可控性和局部性优势。

Abstract: Most recent 3D reconstruction and editing systems operate on implicit and explicit representations such as NeRF, point clouds, or meshes. While these representations enable high-fidelity rendering, they are fundamentally low-level and hard to control programmatically. In contrast, we propose and systematically evaluate a new 3D reconstruction paradigm, 3D Code Synthesis (3D-CoS), where 3D assets are constructed as executable Blender code, a programmatic and interpretable medium. To assess how well current VLMs can use code to represent 3D objects, we evaluate representative open-source and closed-source VLMs in code-based reconstruction under a unified protocol. We further introduce a suite of structured code-synthesis workflows, including blueprint-based planning, Retrieval-Augmented Generation (RAG) over Blender API documentation, few-shot geometric demonstrations, and a component-level Agent workflow for part-wise code generation. To demonstrate the unique advantages of this representation, we further evaluate localized text-driven modifications and compare our code-based edits with a point-cloud-based 3D editing baseline. Our study shows that code as a 3D representation offers strong controllability and locality, yielding stronger edit fidelity and better preservation of unedited regions in our targeted editing evaluation. Our work also analyzes the potential of this paradigm, delineates the current capability frontier of VLMs for programmatic 3D modeling, and highlights code synthesis as a promising direction for editable 3D reconstruction.

[43] GUI-AC: Enhancing Continual Learning in GUI Agents cs.CVPDF

Can Lin, Tao Feng, Hangjie Yuan, Dan Zhang, Yifan Zhu

TL;DR: 本文提出GUI-AC方法，旨在解决GUI智能体在持续学习过程中因界面数据非平稳性（如新领域、新分辨率）导致的性能下降问题。该方法通过引入‘接地确定性’概念，设计了自适应优势估计和动态裁剪两个核心机制，以稳定强化微调过程并增强探索能力。

Details

Motivation: 现有GUI智能体难以泛化到真实世界多样且持续变化的界面环境，其强化微调方法存在奖励不连续、方差高、优势估计噪声大以及探索能力崩溃等问题，阻碍了持续学习。

Result: 大量实验表明，GUI-AC方法在持续学习任务上的性能超越了最先进的基线模型，具体表现为联合改进机制带来的稳定性和性能提升。

Insight: 创新点在于提出了‘接地确定性’来量化模型预测的置信度，并基于此设计了自适应优势估计（抑制噪声，防止策略过度自信）和动态裁剪（放宽裁剪边界，鼓励探索）两个机制，为解决强化微调中的不稳定性和探索不足问题提供了新思路。

Abstract: Graphical User Interfaces (GUIs) serve as the dominant medium for human-computer interaction, yet building GUI agents that generalize across the vast diversity of real-world interface environments, with the same flexibility and robustness that humans naturally exhibit, remains unsolved. Notably, GUI data are inherently non-stationary: the continual emergence of previously unseen interface instances (e.g., novel domains and resolutions) induces persistent distribution shifts, significantly impeding the continual learning of existing GUI agents. Reinforcement fine-tuning (RFT) has attracted considerable attention as a promising approach. Nevertheless, RFT exhibits pronounced instability in its grounding capability, manifested as sharp reward discontinuities and high-variance oscillations. The imbalanced distribution of rollout outcomes introduces substantial noise into advantage estimation, leading to policy overconfidence. The fixed clipping bound suppresses the increase in policy probabilities needed to adapt to new distributions, leading to a collapse in exploration capacity. To address these challenges, we propose GUI-AC, a method that enhances the continual learning capability of GUI agents. GUI-AC introduces grounding certainty to support two core mechanisms: (i) Adaptive Advantage, which down-weights noisy advantage estimates to prevent policy overconfidence; and (ii) Dynamic Clipping, which relaxes the clipping bound to encourage exploration range. Extensive experiments show that these mechanisms jointly improve performance, enabling our method to surpass state-of-the-art baselines. Code is available anonymously at https://anonymous.4open.science/r/GUI-AC.

[44] Audio-Visual Exchange-Aware Token Pruning for Efficient Audio-Visual Captioning cs.CVPDF

Zihan Meng, Dexiang Hong, Weidong Chen, Ziyu Zhou, Bo Hu

TL;DR: 本文提出了一种名为AVEX-Prune的基于强化学习的音视频动态令牌剪枝方法，用于高效的音视频描述生成。该方法通过音视频令牌交换策略，选择真正有价值的令牌，在保留40%令牌的情况下，能在VILA 1.5-8B和VideoLLaMA 2模型上保持与全令牌相当的质量。

Details

Motivation: 音视频描述生成任务中，多模态大语言模型输入的两个模态都会贡献大量令牌，导致预填充自注意力计算呈二次方增长。现有令牌剪枝方法通过硬阈值选择令牌，难以保留真正有价值的令牌，特别是决策边界附近的高混淆令牌。

Result: 在VILA 1.5-8B模型上，40%保留率下CIDEr得分54.5 vs. 全令牌54.6；在VideoLLaMA 2模型上，40%保留率下CIDEr得分57.0 vs. 全令牌56.8，表明在显著减少计算量的同时保持了与全令牌相当的性能。

Insight: 创新点在于提出了音视频令牌交换策略，通过替换低置信度保留令牌为来自同一或另一模态的高置信度候选令牌，并测量描述生成的差异来动态选择真正有价值的令牌，避免了硬阈值选择的局限性。

Abstract: Audio-visual captioning generates natural language descriptions from video and audio content. Multimodal LLMs have advanced this task, but both modalities contribute many tokens to the LLM input, where prefill self-attention scales quadratically. Existing token-pruning methods usually retain tokens by attention, saliency, or cross-entropy loss, yet the hard threshold selection makes it difficult to retain tokens that are truly valuable, especially for high-confusing tokens near the decision boundary. To this end, we propose a AVEX-Prune, an RL-based audio-visual dynamic token pruning method in this work. In our AVEX-Prune, an audio-visual token exchange strategy is proposed to select truly valuable tokens by replacing low-confidence retained tokens with high-confidence candidate tokens from the same or the other modality, and measuring the differences in caption generation from token swaps. AVEX-Prune preserves full-token quality at a 40% retention ratio on both VILA 1.5-8B (54.5 vs. 54.6) and VideoLLaMA 2 (57.0 vs. 56.8).

[45] GRAR: Glass-induced Reflection Artifact Removal in LiDAR Point Clouds cs.CVPDF

Wanpeng Shao, Zeyi Guo, Bo Zhang, Yifei Xue, Tie Ji

TL;DR: 本文提出了一种名为GRAR的统一框架，用于去除LiDAR点云中由玻璃引起的反射伪影。该框架分为两个阶段：第一阶段利用多模态视觉基础模型生成初始玻璃掩膜，并通过几何线索进行精炼和补全；第二阶段提出了一种基于物理的反射感知局部-全局几何相似性描述符，以增强对不完美观测的鲁棒性。

Details

Motivation: 城市环境中采集的地面激光扫描点云常受玻璃反射伪影影响，严重降低下游应用性能。现有方法依赖理想反射对称假设，其性能受限于不准确的玻璃估计和不足的几何表示。

Result: 在多个公共TLS数据集上的大量实验表明，该框架在反射伪影去除方面始终优于最先进的方法。

Insight: 创新点包括：利用多模态基础模型进行高精度玻璃区域估计与补全，以及提出基于实际激光反射几何的物理驱动描述符来联合编码多尺度几何结构和方向一致性，从而提升对不完美观测的鲁棒性。

Abstract: Terrestrial Laser Scanning (TLS) point clouds captured in urban environments frequently suffer from glass-induced reflection artifacts, severely degrading downstream applications. Existing reflection artifact removal methods generally rely on ideal reflection symmetry assumptions, yet their performance is limited by inaccurate glass estimation and insufficient geometric representations. To address these issues, we propose a novel unified framework aimed at robust reflection artifact removal: In the first stage, we leverage a multi-modal vision foundation model to produce initial glass masks, which are then refined using geometric cues to achieve high-precision glass regions, followed by glass completion to recover missing regions caused by no-return measurements on transparent surfaces; In the second stage, we propose a physics-driven descriptor, termed Reflection-aware Local-Global Geometric Similarity (RE-LGGS), which is grounded in actual laser reflection geometry and jointly encodes multi-scale geometric structures and orientation consistency using PCA-based local shape representations, thereby significantly improving robustness against imperfect observations. Extensive experiments on multiple public TLS datasets demonstrate that our framework consistently outperforms state-of-the-art methods in reflection artifacts removal.

[46] PrismAvatar: Pseudo-Multiview Reconstruction and Subpixel Prism Rendering for Real-Time Stereoscopic Communication cs.CV | cs.GRPDF

Chufeng Fang, Dongdong Teng, Lilin Liu

TL;DR: PrismAvatar是一个用于实时立体视频通信的高斯人头化身系统，它通过单目肖像视频重建可控头部化身，并针对无眼镜光栅显示器的横向观看区域进行优化。该方法利用自然头部转动作为伪多视角监督，以约束单目训练中弱观测区域（如头发、耳朵、下颌轮廓和颈部边界），并通过轮廓感知损失和分阶段正则化减少伪影。系统实时渲染32个虚拟视图，并将其编码为4K光栅图像，原型系统在实时跟踪下达到10.65 FPS，经特定对象蒸馏驱动后提升至38.49 FPS。

Details

Motivation: 解决实时立体视频通信中依赖专业捕捉设备或仅提供单一肖像视图的局限性，旨在实现基于单目捕捉的无眼镜立体通信系统。

Result: 在无眼镜光栅显示器上实现实时立体渲染，原型系统在实时跟踪下达到10.65 FPS，经蒸馏优化后提升至38.49 FPS，展示了可行的实时性能。

Insight: 创新点包括利用自然头部转动作为伪多视角监督来增强单目重建的侧视图细节，以及通过子像素编码和光栅显示技术实现无眼镜立体渲染；客观分析认为其将高斯化身与显示硬件特性结合，优化了实时通信的视觉质量和效率。

Abstract: Real-time stereoscopic video communication has long been a goal of immersive telepresence, yet practical systems still require specialized capture rigs or reduce remote users to a single portrait view. We present PrismAvatar, a Gaussian head-avatar system that connects monocular avatar capture with subpixel-encoded glasses-free lenticular display for real-time autostereoscopic communication. From a monocular portrait video, PrismAvatar reconstructs a controllable head avatar and optimizes it for the lateral viewing zones induced by the display. The method uses natural head turns as pseudo-multiview (PMV) supervision to constrain regions that are otherwise weakly observed in monocular training, including hair, ears, jaw contours, and neck boundaries. Reliable side frames are yaw-binned, aligned to virtual cameras, and supervised within a strict head-and-hair domain; contour-aware losses and staged regularization further suppress ghosting, alpha leakage, and depth instability while preserving lateral detail. At runtime, PrismAvatar renders 32 virtual views and encodes them into a 4K lenticular raster with calibrated subpixel-routing masks. The live-tracker prototype sustains 10.65 FPS, and a subject-specific distilled driver raises the same display pipeline to 38.49 FPS.

[47] Improving Adversarial Transferability on Vision-Language Pre-training Models via Surrogate-Specific Bias Correction cs.CV | cs.AI | cs.CRPDF

Lijia Yu, Jiuxin Cao, Yuchen Qiang, Changhao Chen, Yifei Huang

TL;DR: 本文提出DeBias-Attack方法，通过校正对抗优化过程中的代理模型特定偏差来提升对抗样本在视觉语言预训练模型间的可迁移性。该方法维护两个扰动分支：主分支在原始图像上优化扰动以破坏图文对齐，参考分支在弱语义图像上优化扰动以估计代理模型偏差，并通过梯度投影校正来减少偏差，从而增强跨模型攻击效果。

Details

Motivation: 现有基于迁移的黑盒攻击方法严重依赖代理模型，导致对抗优化方向过度拟合代理模型的响应而非输入语义，这种代理模型特定偏差降低了对抗样本在未见目标模型上的可迁移性。

Result: 实验表明，DeBias-Attack在多种VLP模型、下游任务以及开源和闭源的多模态大语言模型上均表现出强大的攻击性能，显著提升了对抗样本的跨模型迁移能力。

Insight: 创新点在于首次通过梯度校正来减少对抗优化中的代理模型特定偏差，具体通过设计弱语义图像分支估计偏差并移除主梯度在参考梯度上的对齐投影；该方法为提升基于迁移的黑盒攻击在复杂多模态模型上的泛化性提供了新思路。

Abstract: Adversarial examples reveal vulnerabilities in Vision-Language Pre-training (VLP) models and provide insights for improving robustness. A key property is cross-model transferability, which enables transfer-based black-box attacks. However, existing attacks often rely heavily on the surrogate model, causing cross-model performance drops. One reason is that adversarial optimization may follow surrogate model responses more than input semantics, making the update direction effective on the surrogate but less transferable to unseen targets. We refer to this dependency as surrogate-specific bias. Motivated by this observation, DeBias-Attack improves transferability by correcting surrogate-specific bias in adversarial optimization directions. It maintains two perturbation branches. The main branch optimizes a perturbation on the original image and obtains the adversarial gradient used to disrupt image-text alignment. The reference branch optimizes a perturbation on a weak-semantic image constructed from the dataset mean image with small Gaussian noise resampled at each iteration. Since this weak-semantic image contains little clear visual content, its optimization reflects surrogate responses more than image semantics, and its reference gradient estimates surrogate-specific bias. DeBias-Attack removes the aligned projection of the main gradient on the reference gradient before updating the adversarial image, then performs context-aware text substitution using the updated adversarial image. DeBias-Attack is the first transfer-based VLP attack that corrects surrogate-specific bias through gradient correction. Experiments show strong performance across VLP models, downstream tasks, and open-source and closed-source multimodal large language models.

[48] Segment and Select: Vision-Language Segmentation in 3D Scenarios cs.CVPDF

Yulin Chen, Zhihang Zhong, Yuenan Hou

TL;DR: 本文提出SEGA3D方法，用于解决3D视觉语言分割任务。该方法摒弃了传统依赖粗糙超点表示的做法，转而采用细粒度掩码候选生成器，并结合大型语言模型（LLM）和语义-空间选择器（SSS）来生成高质量的分割结果，最后通过循环验证模块（LVM）输出最终掩码。

Details

Motivation: 现有3D视觉语言分割方法严重依赖粗糙的超点表示以降低计算复杂度，但这导致分割质量差、物体边界混乱。本文旨在摆脱对超点的依赖，直接利用细粒度视觉信息来提升分割精度。

Result: 在ScanRefer、ScanNet和Matterport3D基准测试上取得了有竞争力的性能。具体而言，在ScanNet和Matterport3D上分别超越了当前最佳方法8.3 mIoU和5.3 mIoU，达到了SOTA水平。

Insight: 创新点在于提出了“分割与选择”（SEGA3D）范式，核心是细粒度掩码候选生成器与LLM驱动的语义-空间选择器的结合。该方法摆脱了对超点表示的依赖，通过端到端的细粒度处理显著提升了分割质量，其循环验证机制也增强了结果的可靠性。

Abstract: 3D vision-language segmentation aims to segment target objects in 3D scenarios according to the linguistic instructions and visual observations. Prior art heavily relies on the coarse superpoint representation to reduce the computation complexity, which suffers from poor segmentation quality and messy object boundaries. In this paper, we propose the SEGment-And-select (SEGA3D) paradigm for 3D visionlanguage segmentation that directly operates on the fine-grained visual information and is free from the superpoint dependency. Specifically, we first leverage a mask candidate generator to provide fine-grained categorical mask candidates, substantially improving the quality of candidate masks over the superpoint counterparts. Then, a Large Language Model (LLM) is utilized to generate the semantic and spatial information based on the linguistic description and visual features. The LLM output and visual features are fed to the Semantic-Spatial Selector (SSS) to produce the top-ranking mask candidates. Eventually, the Loopback Verification Module (LVM) is designed to yield the segmentation mask from the selected candidate masks. Our SEGA3D attains competitive performance on ScanRefer, ScanNet and Matterport3D benchmarks. Notably, our SEGA3D surpasses the top-performing counterpart by 8.3 mIoU and 5.3 mIoU on ScanNet and Matterport3D, respectively. Codes will be available upon publication.

[49] GaussTrace: Provenance Analysis of 3D Gaussian Splatting Models with Evidence-based LLM Reasoning cs.CVPDF

Haoliang Han, Ziyuan Luo, Renjie Wan

TL;DR: 本文提出了GaussTrace，一个用于构建3D高斯溅射（3DGS）模型有向溯源图的新框架。该框架将溯源分析视为一个基于证据的推理问题，通过分析3DGS参数的属性统计特征和进行假设驱动的编辑模拟来生成证据，并利用大型语言模型（LLM）进行结构化思维链推理，从而推断模型间的演化关系。

Details

Motivation: 针对3D高斯溅射模型在数字平台上广泛分享和迭代修改所带来的知识产权保护和法证溯源挑战，需要一种能够有效追踪模型演化历史的方法。

Result: 实验结果表明，GaussTrace能够有效构建不同3DGS模型之间的演化关系，生成准确、可解释且鲁棒的溯源图，且无需模型训练或访问编辑历史。

Insight: 创新点在于将3DGS模型的溯源问题形式化为基于证据的LLM推理任务，结合了属性统计分析和假设驱动模拟来生成证据，并利用LLM的结构化思维链进行可解释的推理，为3D内容溯源提供了一种无需训练、不依赖历史记录的新方法。

Abstract: 3D Gaussian Splatting (3DGS) is a powerful technique for creating high-fidelity 3D assets. However, the widespread sharing and iterative modification of 3DGS models across digital platforms create pressing challenges for intellectual property protection and forensic traceability. To address this, we propose GaussTrace, a novel framework for constructing directed provenance graphs for 3DGS models. GaussTrace formulates provenance analysis as an evidence-based reasoning problem. It builds upon attribute-wise statistical profiling of 3DGS parameters to capture intrinsic properties. Moreover, we introduce hypothesis-driven editing simulations of common operations to provide auxiliary evidence for plausible transformation pathways. These statistical and simulated cues jointly enable a Large Language Model (LLM) to perform structured Chain-of-Thought (CoT) reasoning, yielding directional provenance inferences and explainable edge reasons. Experimental results demonstrate that GaussTrace effectively constructs evolutionary relationships among diverse 3DGS models, delivering accurate, interpretable, and robust provenance graphs without requiring model training or access to editing histories. Project page: https://haolianghan.github.io/GaussTrace.

[50] Globally Localizing Lunar Rover in Pixels via Graph Alignment cs.CVPDF

Mao Chen, Xu Yang, Chuankai Liu, Xiangkai Zhang, Xiaoxue Wang

TL;DR: 本文提出了一种名为WARG（Warped Alignment of Reprojected Graphs）的框架，用于解决月球漫游车在缺乏GNSS信号环境下的全局精确定位问题。该框架通过统一图学习和重投影图匹配，实现了漫游车视角与卫星视角图像之间的鲁棒跨视图对齐。

Details

Motivation: 月球自主探索需要精确的漫游车定位，但GNSS信号的缺失以及局部定位方法的累积漂移严重限制了长距离任务。跨视图定位通过匹配漫游车视图和卫星视图图像，提供了一种无漂移的全局解决方案，但月球环境带来了实体纠缠、视点差异和仿真到真实域偏移等独特挑战。

Result: 在合成数据集LuSNAR上预训练后，WARG的平均测试误差为0.32米，并在合成月球南极区域实现了零样本泛化，误差为3.63米。在玉兔二号漫游车的真实数据上验证时，其在100米×100米搜索区域内的定位误差为1.68米，相当于在1.40米/像素的低分辨率卫星图像中达到近一个像素的精度。模型参数量仅1.56M，是先前轻量级模型的16.12%，在NVIDIA RTX A6000 GPU上运行频率为5.49 Hz。

Insight: 创新点在于提出了一个结合统一图学习和重投影图匹配的框架，以应对月球跨视图定位中的独特对齐挑战。该框架不仅定位精度高、计算高效，还通过跨视图定位学习自然地发展了低层空间感知能力（如语义分割和结构推理），展示了以最小标注成本实现空间智能的潜力。

Abstract: Precise rover localization is a prerequisite for autonomous lunar exploration, yet the absence of Global Navigation Satellite System (GNSS) signals and the cumulative drift of local localization methods severely constrain long-range missions. Cross-view localization provides a promising drift-free global solution by matching rover-view and satellite-view imagery. However, the lunar environment poses unique challenges for correspondence alignment, including inter-entity entanglement, inter-viewpoint divergence, and simulation-to-real domain shift. To address these challenges, we propose Warped Alignment of Reprojected Graphs (WARG), a framework that leverages unified graph learning and reprojected graph matching for robust cross-view alignment. Pretrained on the synthetic LuSNAR dataset, WARG achieves an average test error of 0.32 m and demonstrates robust zero-shot generalization to the synthetic lunar south pole region with an error of 3.63 m. More importantly, when validated on real-world data from the YuTu-2 rover, WARG achieves a localization error of 1.68 m within a 100 m x 100 m search area, corresponding to nearly one-pixel precision in low-resolution satellite imagery with a spatial resolution of 1.40 m/pixel. Beyond accuracy, WARG is computationally efficient, containing only 1.56M parameters, corresponding to 16.12% of previous lightweight models, and operating at 5.49 Hz on an NVIDIA RTX A6000 GPU, approaching GNSS-level update frequency. Finally, we observe that WARG naturally develops low-level spatial awareness, including semantic segmentation and structural reasoning, through cross-view localization learning, highlighting its potential as a promising paradigm for spatial intelligence with minimal annotation cost. The source code is available at https://github.com/maochen-casia/warg.

[51] Can Image Models Imagine Time? ImageTime: A Novel Benchmark for Probing Visual World Modeling Through Spatiotemporal Consistency cs.CV | cs.AIPDF

Xinrui Wu, Lichen Huang

TL;DR: 本文提出了ImageTime基准测试，用于评估图像生成模型在保持时空一致性方面的能力。该基准要求模型根据动作指令生成包含四个有序关键状态（初始状态、动作开始、过渡状态和最终状态）的图像，以诊断模型对视觉世界随时间变化的建模能力。

Details

Motivation: 现有图像生成模型能生成高质量静态图像，但其表示视觉世界随时间变化的能力尚不明确，而实际应用（如故事板、分步插图等）需要模型在多个视觉状态间保持一致性。现有评估主要关注单图像正确性或视频质量，缺乏对时间有序过程连贯性的测评。

Result: 通过多模型家族基准测试，ImageTime揭示了当前图像生成系统在保持连贯视觉世界状态方面的成功、失败和偏差情况，并利用GPT-5.5在结构化VLM-as-judge协议下生成可解释的能力分数、诊断子分数和失败标签。

Insight: 创新点在于设计了四关键帧协议，比单图像生成更具时间要求，同时避免了密集视频动态的混淆；基准采用渐进能力层次结构，将场景分解为阶段状态谓词、跨帧时间约束和禁止因果违规，为评估视觉世界建模提供了新方法。

Abstract: Image generation models now produce high-quality static images, yet their ability to represent how a visual world changes over time remains poorly understood. Practical workflows such as storyboarding, step-by-step illustration, reference-guided editing, and video previsualization require models to preserve identities, objects, spatial relations, and causal order across multiple visual states. Existing evaluations largely measure single-image correctness, compositional alignment, or video quality, leaving open whether an image model can coherently imagine a temporally ordered process. We introduce ImageTime, a diagnostic benchmark that uses spatiotemporal consistency as a behavioral probe of visual world modeling in image generation. Given an action instruction, and optionally a reference image specifying the initial state, a model must generate one image containing four ordered key states: initial state, action onset, transition state, and final state. This four-keyframe protocol is more temporally demanding than single-image generation while avoiding the confounds of dense video dynamics. ImageTime organizes tasks with a progressive capability hierarchy and decomposes each scenario into stage-wise state predicates, cross-frame temporal constraints, and forbidden causal violations. GPT-5.5 scores all generated images under a structured VLM-as-judge protocol, producing interpretable capability scores, diagnostic subscores, and failure labels. Through multi-family benchmarking, ImageTime reveals where current image generation systems succeed, fail, and drift when asked to maintain coherent visual world states over time.

[52] ManiSplat: Manipulation Trajectory Synthesis from Monocular Video via Decoupled 3D Gaussian Splatting cs.CVPDF

Wenhao Hu, Haonan Zhou, Liu Liu, Yun Du, Xinjie Wang

TL;DR: ManiSplat是一个从单目机器人视频重建可控、解耦的3D高斯数字孪生的统一框架。它通过图结构解耦表示将机器人、物体和背景分离为独立优化的高斯子场，并利用任务导向的时空对齐模块构建准确的伪真值轨迹，最终通过联合光度-几何优化确保重建场景的时空一致性和物理一致性。

Details

Motivation: 解决从真实世界观察中重建动态交互式3D场景的挑战，特别是针对包含铰接式机器人和可操纵物体的交互环境，这些环境因复杂的接触交互和突然的位姿变化而难以处理。

Result: 大量实验表明，该方法能够以高保真度和可控性重建交互驱动的动态场景，有效支持下游机器人任务和策略学习。

Insight: 创新点在于提出了图结构解耦表示以实现场景组件的独立优化，以及任务导向的时空对齐模块，该模块利用操作任务的内在逻辑（在运动和技能阶段之间交替）来构建伪真值轨迹，从而确保重建的稳定性和准确性。

Abstract: Reconstructing dynamic and interactive 3D scenes from real-world observations remains a fundamental challenge in computer vision and robotics. While recent advances in 3D Gaussian Splatting have enabled high-fidelity static reconstruction, extending it to interactive environments with articulated robots and manipulable objects remains difficult due to complex contact interactions and abrupt pose changes. To address these challenges, we introduce ManiSplat, a unified framework that reconstructs controllable and decoupled Gaussian digital twins directly from monocular ego-view robotic videos. Our method introduces a Graph-Structured Disentangled Representation that separates the robot, objects, and background into independently optimizable Gaussian subfields organized within a scene graph. To ensure stability, we propose a Task-Oriented Spatio-Temporal Alignment module that leverages the inherent logic of manipulation tasks-alternating between Motion and Skill phases-to construct accurate pseudo-ground-truth trajectories. Finally, a joint photometric-geometric optimization ensures the reconstructed scenes are temporally coherent, physically consistent, and simulation-ready. Extensive experiments demonstrate that our approach reconstructs interaction-driven dynamic scenes with high fidelity and controllability, effectively supporting downstream robotic tasks and policy learning.

[53] Envision4D: Envisioning Visual Futures via Feed-forward 4D Gaussian Splatting for Autonomous Driving cs.CVPDF

Qi Song, Yifei He, Chi Zhang, Zheng Fu, Xuhe Zhao

TL;DR: 本文提出了Envision4D，一个用于自动驾驶场景中无姿态未来外推的全自监督前馈框架。该方法通过未来姿态预测模块、层内时序注意力、条件化运动提升以及渐进式训练策略，解决了现有前馈方法在大位移下产生重影、依赖简化运动假设或严格未来先验的问题，实现了高质量的未来视图合成。

Details

Motivation: 现有用于动态场景预测的前馈范式主要针对插值设计，当扩展到未来外推时，在大位移下会产生重影伪影，并且受限于简化的运动假设或严格的未来先验。本文旨在克服这些挑战，实现无姿态、自监督的未来场景演化预测。

Result: 大量实验表明，Envision4D在未来的视图合成任务中取得了最先进的性能，显著优于现有方法。

Insight: 创新点包括：1) 通过迭代去噪过程推断未来相机参数的未来姿态预测模块；2) 用于捕捉非线性动态的层内时序注意力和条件化运动提升，将高度不确定的外推过程转化为稳健的关系映射；3) 用于稳定无监督运动学习、对抗误差累积的渐进式训练策略。

Abstract: Forecasting the future evolution of dynamic scenes is crucial in autonomous driving. However, existing feed-forward paradigms are primarily designed for interpolation. When extended to future extrapolation, they suffer from ghosting artifacts under large displacements and are constrained by simplified motion assumptions or strict future priors. To overcome these challenges, we propose Envision4D, a fully self-supervised feed-forward framework for pose-free future extrapolation. Specifically, we introduce a Future Pose Prediction module that infers future camera parameters via an iterative denoising process. Furthermore, to capture non-linear dynamics, we propose In-layer Temporal Attention and employ Conditioned Motion Lifting, which transforms the highly uncertain extrapolation process into robust relational mappings. Finally, a Progressive Training Strategy is utilized to stabilize unsupervised motion learning against error accumulation. Extensive experiments demonstrate that Envision4D achieves state-of-the-art performance, significantly outperforming existing methods in future view synthesis.

[54] Kwai Keye-VL-2.0 Technical Report cs.CVPDF

Kwai Keye Team, Bin Wen, Changyi Liu, Chengru Song, Chongling Rao

TL;DR: 该论文介绍了Kwai Keye-VL-2.0-30B-A3B，一个开源的混合专家（MoE）多模态基础模型，旨在推进长视频理解和智能体能力。该模型通过采用DeepSeek稀疏注意力机制处理256K上下文，并引入跨模态多教师策略蒸馏方法，在仅激活30亿参数的情况下，实现了在视频理解、时序定位、推理和智能体基准测试中的SOTA性能。

Details

Motivation: 为了解决小时级别长视频理解中存在的超长上下文、信息冗余和计算成本高昂的挑战，并克服多任务对齐过程中的灾难性遗忘问题，从而赋能跨代码、工具和搜索场景的先进智能体协作。

Result: 在视频理解（Video-MME-v2, LongVideoBench）、时序定位（TimeLens）、推理和STEM等多个基准测试中，Keye-VL-2.0-30B-A3B在同等规模模型中达到了最先进的（SOTA）性能，尤其在TimeLens的细粒度时序定位和长视频理解任务上表现出色。

Insight: 主要创新点包括：1）首次将DeepSeek稀疏注意力（DSA）适配到基于GQA的多模态架构中，实现无损的256K长上下文处理；2）提出跨模态多教师策略蒸馏（MOPD）方法，结合Context-RL和Video-RL，将密集的令牌级教师反馈蒸馏回MoE骨干网络，有效缓解灾难性遗忘并赋能多模态自我纠正的智能体协作。

Abstract: We introduce Kwai Keye-VL-2.0-30B-A3B, an open-source Mixture-of-Experts (MoE) multimodal foundation model designed to advance long-video understanding and agentic intelligence. To address the challenges of ultra-long contexts, information redundancy, and prohibitive computational costs inherent in hour-level videos, Keye-VL-2.0 is the first to adapt DeepSeek Sparse Attention (DSA) to GQA-based multimodal architectures, enabling lossless 256K context processing while capturing critical frames and long-range temporal dependencies. This architecture is underpinned by a highly optimized training and inference infrastructure, including scalable video I/O, heterogeneous ViT-LM parallelism, and custom DSA kernels that significantly maximize throughput and minimize computational overhead. Furthermore, to overcome the algorithmic dilemma of catastrophic forgetting during multi-task alignment, we introduce Cross-Modal Multi-Teacher On-Policy Distillation (MOPD) paired with Context-RL and Video-RL. By distilling dense token-level teacher feedback from on-policy rollouts back into the MoE backbone, which activates only 3B parameters, Keye-VL-2.0 natively empowers advanced agent collaboration across Code, Tool, and Search scenarios with multimodal self-correction. Extensive evaluations across video understanding, temporal grounding, reasoning, STEM, and agent benchmarks demonstrate that Keye-VL-2.0-30B-A3B achieves state-of-the-art performance among models of similar scale, particularly excelling in fine-grained temporal localization on TimeLens and long-video comprehension on Video-MME-v2 and LongVideoBench. We release our model checkpoints to accelerate community progress toward scalable and robust multimodal agentic applications.

[55] Analyzing Training-Free Corruption Detection for Object Detection Datasets cs.CV | cs.DBPDF

Christian Sieberichs, Simon Geerkens, Thomas Waschulzik, Viswanathan Ramesh, Alexander Braun

TL;DR: 本文分析了基于特征空间的无训练方法在目标检测数据集中检测标注错误的适用性。通过调整现有方法，研究发现该方法能可靠地暴露语义错标，但难以检测位置错误。研究在VOC2012和KITTI数据集上，使用多种预训练嵌入模型和合成噪声类型（对称、非对称和位置噪声）以及真实标注错误进行了评估。

Details

Motivation: 计算机视觉数据集中的标注错误普遍存在，会显著降低训练系统的性能，尤其是在目标检测等复杂任务中。现有方法（包括无训练的特征空间方法）虽能快速、可解释地分析标注，但其在包含语义和空间信息的目标检测标注上的行为仍未被充分探索。

Result: 研究在VOC2012和KITTI数据集上评估了方法。结果表明，基于特征空间的方法能可靠地检测语义错标，但对位置错误的检测效果不佳。

Insight: 创新点在于将特征空间方法专门应用于目标检测标注错误检测，并系统分析了其对语义与位置错误的检测能力差异。客观来看，该研究揭示了无训练方法在复杂标注任务中的局限性，为未来开发更全面的错误检测工具提供了重要见解。

Abstract: Annotation errors are widespread in computer vision datasets and can significantly degrade the performance of systems trained on them, particularly in complex tasks such as object detection. Several approaches exist to identify annotation errors, including training-free feature-space methods which provide a fast and interpretable way to analyze annotations. However, the behavior on object detection annotations, which include semantic and spatial information, remains largely unexplored. In this work we analyze the applicability of feature-space-based approaches for detecting annotation errors in object detection datasets. By adapting an existing feature-space method, we show that such approaches reliably expose semantic mislabel, while positional errors remain difficult to detect. We evaluate this behavior across multiple pretrained embedding models, synthetic noise types (symmetric, asymmetric, and positional), and real-world annotation errors using VOC2012 and KITTI. All code and real-world corruptions are publicly available at the following repository: https://github.com/ ChristianSieberichs/BoundingBox_corruption_detection

[56] FadeMem: Distance-Aware Memory Consolidation for Autoregressive Video Diffusion cs.CVPDF

Yu Lu, Junjie Yang, Piotr Koniusz, YuXin Song, Yi Yang

TL;DR: 本文提出FadeMem，一种用于自回归视频扩散模型的距离感知KV内存整合机制。该方法在固定缓存预算下，将历史KV块组织成时间层次结构，根据时间距离动态分配内存粒度，以解决长视频生成中历史KV缓存随视频长度增长的问题。

Details

Motivation: 自回归视频生成器通过生成连续的时间片段来合成长视频，但其历史KV缓存会随视频长度增长，现有有界缓存方法（如局部窗口、汇令牌或压缩内存状态）通常为历史的不同部分分配固定角色，无法适应不同时间距离的信息重要性差异。

Result: 实验表明，与现有的有界缓存策略相比，FadeMem在主体一致性、背景稳定性和时间连贯性方面均有提升，具体定量结果未在摘要中提供，但暗示了优于现有方法的性能。

Insight: 核心创新在于提出了基于幂律时间分配策略的动态内存整合机制，其动机源于时间衰减的频率依赖性：精细细节快速去相关，而粗略场景结构和身份信息在更长时间范围内保持有用。该方法无需改变模型架构，即可在单一缓存内实现“近密远疏”的内存组织，兼顾短期动态的近期上下文和长期身份/场景连贯性的紧凑锚点。

Abstract: Autoregressive video generators synthesize long videos by generating successive temporal segments, but their historical KV cache grows with video length. Existing bounded-cache methods reduce this cost with local windows, sink tokens, or compressed memory states, yet they usually assign fixed roles to different parts of the history. We propose FadeMem, a distance-aware KV memory consolidation mechanism that organizes historical KV blocks into a temporal hierarchy under a fixed cache budget. This design is motivated by frequency-dependent temporal decay: fine details decorrelate quickly, while coarse scene structure and identity remain useful over longer horizons. During generation, new history is inserted as fine-grained entries, while older adjacent entries are progressively merged under a power-law temporal allocation schedule, yielding a dense-near, sparse-far memory within one cache. Without architectural changes, FadeMem preserves recent context for short-term dynamics and compact long-range anchors for identity and scene coherence. Experiments show improved subject consistency, background stability, and temporal coherence over existing bounded-cache strategies.

[57] Vector Map as Language: Toward Unified Remote Sensing Vector Mapping cs.CVPDF

Yinglong Yan, Yunkai Yang, Haoyi Wang, Wei Fu, Linshan Wu

TL;DR: 本文提出了一种名为VecLang的统一范式，将多类别遥感矢量地图生成重新定义为结构化文本生成任务。该方法将地理空间实体的几何、语义和拓扑信息编码为类似GeoJSON的矢量语言，并通过渐进式视觉-语言映射框架和分层优化策略来可靠地生成这种语言。

Details

Motivation: 现有矢量地图生成方法通常将对象表示为多边形或图结构，前者难以捕捉拓扑关系，后者则容易模糊实例边界，且都只适用于特定类别。本文观察到自然语言作为一种灵活的表达媒介，能够统一容纳异构的地图元素，从而提出用语言来统一表示矢量地图的动机。

Result: 在构建的包含5.4万张图像和80万个实例的VecMap-Bench数据集上进行的大量实验表明，VecLang能够同时处理单类别和多类别矢量地图生成任务，并在跨数据集和开放词汇泛化方面取得了强大的性能。

Insight: 核心创新点在于将矢量地图生成视为一种语言生成任务，提出了‘矢量地图即语言’的统一范式。这包括设计一种GeoJSON-like的矢量语言来表示异构地图元素，以及一个包含定位和生成阶段的渐进式框架，并引入基于强化学习的层次化优化来提升生成质量。这种范式转换为实现跨类别的统一建模提供了新思路。

Abstract: Remote sensing vector mapping aims to generate structured maps of geospatial entities, such as buildings, roads, and water bodies, from remote sensing imagery. In practice, vector maps usually contain multiple category layers and heterogeneous entity structures, requiring a unified model for diverse mapping needs. However, existing methods typically represent vector objects as polygons or graphs, making them suitable only for specific categories: polygons poorly capture topological relations, while graphs often blur instance boundaries. We observe that language, as a natural medium for human communication, offers a flexible and expressive representation that can accommodate heterogeneous map elements, including geometry, semantics, and topolog. Motivated by this insight, we propose Vector Map as Language (VecLang), a unified paradigm that reformulates multiclass vector mapping as structured text generation. VecLang encodes the common elements of different geospatial entities into a GeoJSON-like vector language, enabling cross-category modeling within a shared textual format. To generate this language reliably, we design a progressive vision-language mapping framework that first localizes vectorization units and then generates structured map elements. We further introduce Hierarchical Vector Language Optimization, which uses reinforcement learning to improve syntax validity, content fidelity, and map executability. We also build VecMap-Bench with 54K images and 800K instances, supporting training and evaluation across standard and generalization settings. Extensive experiments demonstrate that VecLang handles both single-class and multiclass vector mapping while achieving strong cross-dataset and open-vocabulary generalization. The model and dataset are publicly available at https://github.com/yyyyll0ss/VecLang.

[58] Patient-Level Diagnosis of Acute Myeloid Leukemia via Deep Learning Analysis of Bone Marrow Smear cs.CV | physics.med-phPDF

Yuqi Ma, Tianyi Wang, Weihua Meng, Hongru Chen, Fajin Tao

TL;DR: 本文提出了一种基于深度学习的细胞到患者级诊断流程，用于从骨髓涂片图像中辅助诊断急性髓系白血病（AML）。该方法通过YOLO模块分割细胞，利用EfficientNet-B0分类器识别专家定义的复合母细胞样细胞（CBLC）类别，并将细胞级预测汇总为患者级CBLC比例，以支持AML诊断。

Details

Motivation: 骨髓涂片检查对AML评估至关重要，但人工单细胞解读耗时耗力，且患者级诊断需要整合大量细胞观察。本研究旨在开发一个自动化深度学习流程，以减轻人工负担并提高诊断效率。

Result: 在包含258名患者的多中心数据上，该方法在内部验证中表现稳定，并在外部验证队列（来自三个中心的89名患者）中保持了良好的泛化能力，在中心4、5、6上的集成加权F1分数分别为0.9076、0.8696和0.9124。

Insight: 创新点包括：1) 采用专家定义的复合母细胞样细胞（CBLC）类别作为分类目标，而非严格识别AML原始细胞，这更贴合临床实际；2) 通过两阶段训练策略（GT-to-YOLO和YOLO-to-YOLO）结合类别不平衡校正、中心-边界正则化和形态学辅助监督，提升了模型性能；3) 将细胞级预测聚合为患者级诊断指标，实现了从微观细胞分析到宏观患者诊断的端到端流程。

Abstract: Bone marrow smear review remains important for acute myeloid leukemia (AML) assessment, but manual single-cell interpretation is labor-intensive and patient-level diagnosis requires aggregation of many cellular observations. We present a cell-to-patient deep learning pipeline for AML-assisted diagnosis from bone marrow smear images. The study included 258 patients from six anonymized centers, including a main cohort of 169 patients from Centers 1-3 and an external validation cohort of 89 patients from Centers 4-6. A 16-category cell annotation vocabulary was used to describe the global cellular composition, including granulocytic, monocytic, erythroid, lymphoid, eosinophilic, and other cells. Rather than identifying strict AML blasts or leukemic blasts, the model targets an expert-defined composite category termed Composite Blast-like Cells (CBLC), comprising N, N1, M, M1, R, R1, J, and J1 according to the project-wide morphological standard. A fixed YOLO-based segmentation module detected cells, predicted contours were matched to expert polygon annotations by contour IoU, and standardized single-cell crops were generated. An EfficientNet-B0 classifier was trained through a two-stage GT-to-YOLO and YOLO-to-YOLO strategy with class-imbalance correction, center-border regularization, and morphology-assisted supervision. Cell-level predictions were aggregated into patient-level CBLC ratios for AML-oriented diagnostic support. The pipeline achieved stable internal validation and maintained external generalization, with ensemble weighted F1-scores of 0.9076, 0.8696, and 0.9124 on Centers 4, 5, and 6, respectively.

[59] A Multimodal RGB and Events Dataset for Hand Detection in First-Person View cs.CVPDF

Bharghav Kota, Yulia Sandamirskaya

TL;DR: 本文提出了一种从现有RGB数据集合成事件相机数据的方法，并构建了一个用于第一人称视角手部检测的多模态数据集。该数据集包含RGB和事件流数据，通过调整合成参数模拟不同光照和尺度条件，并使用微调的YOLOv8模型生成标注。实验表明，基于该数据集的多模态检测方法能达到与现有先进方法相当的性能。

Details

Motivation: 解决在移动机器人系统中，传统相机因运动模糊（尤其在暗光条件下）导致手部检测率受限的问题，同时缓解事件相机在目标检测任务中训练数据不足的瓶颈。

Result: 使用所构建的多模态数据集，结合RGB和事件相机进行手部检测，其性能与当前最先进方法（SOTA）相当。

Insight: 创新点在于提出了一种从现有RGB数据集（Egohands）合成事件相机数据的方法论和工具链（v2e），并构建了首个面向第一人称视角手部检测的多模态（RGB+事件）合成数据集，为事件相机在视觉任务中的应用提供了数据基础。

Abstract: Existing hand detection algorithms work on images and the detection rate is restricted by the frame rate of the camera. In hand detection applications for moving robotic systems, conventional cameras cause motion blur, especially in darker lighting conditions. We can leverage the use of event-based cameras which possess a high dynamic range, high temporal resolution, and low power consumption. Recent work has shown that using a stereo setup of an event-based and a frame-based camera improves detection accuracy and the bandwidth-latency tradeoff. The main bottleneck in using event-based cameras in object detection and recognition tasks is a relatively low amount of training data. In this work, we propose a methodology and an exemplary synthetic event-based hand dataset from an egocentric, first-person view perspective. The data is synthesized from the existing RGB Egohands dataset with the v2e toolbox. Parameters of the v2e toolbox are varied to provide versions of the dataset with different lighting conditions and scales. Ground truth detections are generated with a fine-tuned YOLOv8 model which is applied to the RGB images in the Egohands dataset and interpolated on the high-temporal resolution events. We use the multi-modal dataset to perform hand detection with existing object detection algorithms which use a multi-modal setup of event and RGB cameras and demonstrate performance comparable to the state-of-the-art.

[60] Spatially Selective Self-Training for Unsupervised Building Change Detection cs.CVPDF

Wafaa I. M. Hussin, Zhi Lu, Anas M. I. Mohammed, Xiang Zhou, Ratiba A. H. Abubaker

TL;DR: 本文提出了一种名为SST-CD的空间选择性自训练框架，用于解决无监督建筑物变化检测问题。该方法将完全无标签的变化检测重新定义为在噪声伪监督下的端到端检测器学习，通过局部一致性准则筛选可靠像素进行训练，并引入特征适配器和基于原型的解码器来稳定训练过程。

Details

Motivation: 现有无标签方法通常遵循差异到掩码的范式，直接使用时序差异或基础模型响应作为最终变化图，但这些通用时序差异与建筑物定义的结构变化之间存在差距，且易受外观变化、配准误差和非建筑物修改等噪声干扰。

Result: 在LEVIR-CD、WHU-CD和DSIFN-CD三个基准数据集上的实验表明，SST-CD分别取得了83.08%、91.69%和86.60%的F1分数，超越了现有的无监督和无标签基线方法。

Insight: 创新点在于将无监督变化检测重构为噪声伪监督下的检测器学习，并提出了空间选择性自训练机制，通过局部一致性过滤不可靠监督信号；同时，特征适配器和原型解码器的设计有助于学习紧凑的特征表示，提升了模型的鲁棒性。

Abstract: Unsupervised building change detection aims to learn building-change masks from unlabeled bi-temporal remote sensing images. Existing label-free methods often follow a discrepancy-to-mask paradigm, directly using temporal differences, frozen foundation-model responses, prompt-based outputs, or post-processing results as final change maps. Although these strategies provide annotation-free cues, they do not learn a task-specific building-change detector and remain vulnerable to the gap between generic temporal discrepancies and building-defined structural changes. In practice, such discrepancies are often noisy and task-irrelevant, as appearance shifts, registration errors, and non-building modifications can produce strong but misleading responses. To address this problem, we propose SST-CD, a spatially selective self-training framework that reformulates fully label-free building change detection as end-to-end detector learning under noisy pseudo supervision. SST-CD uses temporal discrepancies as candidate pseudo labels and trains the detector only on spatially reliable pixels, whose reliability is estimated by a local consistency criterion that filters inconsistent regions from supervision. To further stabilize noisy self-training, a lightweight feature adapter recalibrates bi-temporal features, while a prototype-based decoder produces compact change and no-change representations. Experiments on LEVIR-CD, WHU-CD, and DSIFN-CD show that SST-CD achieves F1 scores of 83.08%, 91.69%, and 86.60%, respectively, outperforming existing unsupervised and label-free baselines. Code will be made publicly available.

[61] HarmoView: Harmonizing Multi-View Constraints for Identity-Consistent Video Generation cs.CVPDF

Cong Wang, Zhentao Yu, Hongmei Wang, Weicong Liang, Zixiang Zhou

TL;DR: 本文提出了HarmoView框架，用于解决身份一致视频生成中，在大视角变化下难以保持外观保真度的问题。该框架通过多级特征注入、可学习代理令牌和Jump-RoPE等架构改进，结合渐进式视角课程训练策略，有效整合多视角参考输入，并在构建的大规模多视角数据集上验证了其优越性能。

Details

Motivation: 当前身份一致视频生成方法在大视角变化下难以保持外观保真度，而多视角参考输入缺乏有效的整合框架且数据稀缺，因此需要开发一个能稳健处理多视角约束的生成框架。

Result: 在包含52个独特身份、100个手动整理案例的多视角基准测试中，HarmoView显著优于开源基线模型，并与领先的闭源引擎性能相当，达到了身份一致视频生成的SOTA水平。

Insight: 创新点包括：通过多级特征注入提供持久的低级外观锚点以增强身份保持；使用可学习代理令牌统一异构参考布局并解决参考-视角不匹配问题；开发Jump-RoPE进行身份特征隔离以减少串扰；采用渐进式视角课程训练策略稳定过渡到高保真空间推理。

Abstract: Current identity-consistent video generation methods struggle to preserve appearance fidelity under large viewpoint changes. While introducing multi-view reference input offers a natural solution, progress remains constrained by the lack of effective frameworks for multi-view inputs and the scarcity of multi-view data. We address these challenges by proposing HarmoView, a robust framework for identity-consistent video generation that effectively integrates multi-view cues through three architectural refinements complemented by a staged training curriculum. Specifically, we first introduce Multi-level Feature Injection to anchor identity fidelity; by injecting raw ViT features from frontal references alongside text tokens via cross-attention, MFI provides persistent low-level appearance anchors that complement the high-level identity features within DiT blocks, leading to enhanced identity preservation. Then, we employ learnable proxy tokens to unify heterogeneous reference layouts across single-/multi-view settings while simultaneously resolving the reference-view mismatch problem. Jump-RoPE is further developed for identity-wise feature isolation to reduce identity crosstalk. To activate these structural capabilities while preserving the original generative priors, we propose the Progressive View Curriculum. This four-stage training strategy employs view dropout to facilitate a stable transition from vanilla T2V generation to high-fidelity, identity-persistent spatial reasoning. Furthermore, we construct a large-scale multi-view dataset to address the issue of data scarcity. Extensive evaluation on our multi-view benchmark, comprising 100 manually-curated cases spanning 52 unique identities, demonstrates that HarmoView significantly outperforms open-source baselines and matches leading closed-source engines, achieving state-of-the-art performance in identity-consistent video generation.

[62] Earth-OneVision: Extending Remote Sensing Multimodal Large Language Models to More Sensor Modalities and Tasks cs.CV | cs.AIPDF

Miaoxin Cai, Guanqun Wang, Wei Zhang, Guangyao Zhou, Yin Zhuang

TL;DR: 本文提出了Earth-OneVision，一个参数为20亿的遥感多模态大语言模型，旨在统一处理六种传感器模态（光学、SAR、红外、多光谱、时序、视频）及其跨模态融合，覆盖9类任务。模型通过三个创新机制解决了多模态对齐、异构输出统一和复合域差距问题，并在一个包含约3400万问答对的新数据集MMRS-OneVision上训练。

Details

Motivation: 现有遥感多模态大语言模型支持的传感器类型和任务范围狭窄，导致对地球的认知碎片化，且未能充分利用跨模态的地球科学知识。

Result: 在多个基准测试中取得有竞争力或SOTA的结果：在光学视觉定位任务OPT-RSVG上达到87.52% P@0.5；在SAR视觉问答基准SARLANG-Bench上达到80.68%，超过70亿参数模型7%以上；在多光谱分类任务BigEarthNet-MS上达到75.74%召回率；在跨模态推理基准EarthMind-Bench上达到81.94% MCQ准确率。

Insight: 创新点包括：1) 全粒度视觉-语言对齐机制；2) 空间-语言同构序列化机制，统一异构空间输出；3) 渐进式跨模态适应机制，分阶段解决复合域差距。从客观角度看，其统一多传感器模态与任务的单一自回归框架设计，以及为此构建的大规模多模态指令数据集，具有显著的工程与研究价值。

Abstract: RS-MLLMs enable natural-language understanding and spatial reasoning over earth observation imagery. However, existing models support only a narrow range of sensor types and tasks, yielding a fragmented view of the earth and leaving cross-modal geoscientific knowledge largely unexploited. This work presents Earth-OneVision, a 2B RS-MLLM that unifies six sensor modalities (i.e., optical, SAR, infrared, multispectral, temporal, and video) and cross-sensor fusion across 9 task categories within a single autoregressive framework. Three dedicated mechanisms address three bottlenecks. Full-Granularity Vision-Language Alignment (FGVLA) aligns multi-level visual features with the multi-dimensional language space. Spatial-Linguistic Isomorphic Serialization (SLIS) unifies heterogeneous spatial outputs as autoregressive tokens. Progressive Cross-Modality Adaptation (PCMA) decomposes the compound domain gap into sequential stages, tackling the viewpoint and imaging physics gaps in turn. To support joint training, MMRS-OneVision is constructed with ~34M QA pairs spanning all six sensor modalities and cross-sensor fusion across 9 task categories, substantially exceeding existing RS multimodal instruction datasets. With only 2B parameters, Earth-OneVision achieves competitive or state-of-the-art results across extensive benchmarks, consistently matching or outperforming 4B-72B RS-MLLMs. It achieves 87.52% P@0.5 on the OPT-RSVG testset for optical visual grounding and 80.68% on the SAR VQA benchmark SARLANG-Bench, exceeding 7B models by over 7%. It further achieves 75.74% recall on the BigEarthNet-MS testset for multispectral classification, and 81.94% MCQ accuracy on EarthMind-Bench for cross-modality reasoning.

[63] SCAIL-2: Unifying Controlled Character Animation with End-to-end In-Context Conditioning cs.CVPDF

Wenhao Yan, Fengjia Guo, Zhuoyi Yang, Jie Tang

TL;DR: 本文提出了SCAIL-2框架，旨在实现端到端的可控角色动画生成。该方法通过直接将驱动视频与参考序列拼接，并利用上下文掩码条件与模式特定的RoPE作为软指导，绕过了传统方法依赖的中间表示（如姿态骨架），从而减少了信息损失。为了解决端到端数据的缺乏，作者构建了包含异构任务的MotionPair-60K数据集，并提出了Bias-Aware DPO来缓解合成数据在细节区域的误差。

Details

Motivation: 现有可控角色动画方法严重依赖中间表示（如姿态骨架或背景掩码）来表示运动和环境，这不可避免地导致信息丢失。本文旨在绕过这些中间步骤，实现端到端的角色动画，直接从输入视频中获取所有必需的视觉信息。

Result: 广泛的实验表明，该方法在各种角色动画任务中显著优于现有的最先进方法。

Insight: 主要创新点包括：1) 通过上下文掩码条件和模式特定的RoPE实现任务统一与软指导，超越了纯文本指令和原始视觉信息；2) 构建了大规模端到端运动传递数据集MotionPair-60K以解决数据稀缺问题；3) 提出了Bias-Aware DPO偏好对齐方法来缓解合成数据在细节上的误差。这些方法共同实现了无需中间表示的端到端角色动画生成。

Abstract: Controlled character animation requires transferring motion from a driving sequence to a reference character. Prior works heavily rely on intermediate representations, including pose skeletons to represent motion or masked background to represent environment, which inevitably leads to information loss. To address this, we present SCAIL-2, an framework that bypasses those intermediates and achieves \textbf{end-to-end} character animation. By directly concatenating driving videos to the sequence, the model can obtain all the required visual information from the input video. To address lack of end-to-end data, we unify sub-tasks of character animation with decoupled conditions and then curate a pipeline to synthesize MotionPair-60K, an end-to-end motion transfer dataset containing heterogeneous tasks of character animation. To archive the unification, we utilize in-context mask conditioning and mode-specific RoPE as soft guidance beyond textual instructions and raw visual information. To address synthetic discrepancy in detailed regions, we propose Bias-Aware DPO to construct preference items to mitigate the errors. Extensive experiments demonstrate that our method substantially outperforms existing state-of-the-art approaches in various character animation tasks. A large subset of synthetic data as well as model weights will be released at our project page: https://teal024.github.io/SCAIL-2/.

[64] LIBERO-Occ: Evaluating and Improving Vision-Language-Action Models under Scene-Induced Occlusion via Viewpoint Imagination cs.CV | cs.AIPDF

Taishan Li, Jiwen Zhang, Siyuan Wang, Xuanjing Huang, Zhongyu Wei

TL;DR: 本文研究了视觉-语言-动作模型在场景诱导遮挡下的性能挑战，提出了LIBERO-Occ基准来评估模型在部分可观测环境中的表现。针对现有VLA模型在遮挡情况下性能显著下降的问题，作者提出了Viewpoint Imagination方法，通过生成互补视角来增强动作预测的鲁棒性。

Details

Motivation: 现有VLA模型在标准操作基准上表现良好，但假设任务相关物体完全可见，这在现实场景中往往不成立，因为遮挡会导致部分可观测性问题。

Result: 实验表明，最先进的VLA模型在遮挡条件下性能大幅下降。提出的VIM方法在多个任务套件、遮挡类型和严重程度下均提高了鲁棒性，且无需部署额外摄像头。

Insight: 创新点在于将视角想象作为感知补全机制，通过生成互补视图来整合观察和想象证据，为解决部分可观测操作问题提供了新思路。LIBERO-Occ基准的建立也为系统评估VLA模型在遮挡场景下的性能提供了标准。

Abstract: Vision-Language-Action (VLA) models achieve strong performance on standard manipulation benchmarks, but most evaluations assume that task-relevant objects are fully visible. This assumption often fails in realistic settings, where occlusion makes manipulation partially observable. In this paper, we study \textit{scene-induced occlusion} as a fundamental challenge for VLA models and introduce \textbf{LIBERO-Occ}, an occlusion-oriented extension of LIBERO. Experiments show that state-of-the-art VLAs suffer substantial performance degradation under occlusion. To address this issue, we propose \textbf{Viewpoint Imagination (VIM)}, which generates a complementary view from an occluded primary observation and conditions action prediction on both observed and imagined evidence. VIM improves robustness across task suites, occlusion types, and severity levels without requiring additional cameras at deployment time, suggesting that viewpoint imagination is an promising mechanism for perception completion in partially observable manipulation. Our benchmark and corresponding code are available at: \href{https://github.com/litsh/Libero-Occ}{https://github.com/litsh/Libero-Occ}.

[65] Listen, Look, and Learn: Learning Without Forgetting through SAM-Audio cs.CVPDF

Avi Gupta, Nilotpal Sinha, Vishnu Raj, Sambuddha Saha, Pratik Joshi

TL;DR: 本文提出了一种基于SAM-Audio的视听类增量学习方法，通过利用其密集的音频和视觉表征，并采用一种新颖的引导注意力策略，让音频特征在上下文中指导视觉表征，以缓解灾难性遗忘问题。

Details

Motivation: 类增量学习旨在持续学习新类而不遗忘旧知识，但视听模态下的CIL研究不足，且像SAM-Audio这样的基础多模态模型在增量设置中表现不佳，本文旨在填补这一空白。

Result: 在视听CIL基准上的广泛评估表明，该方法在性能上持续优于最先进的方法。

Insight: 创新点在于将SAM-Audio的视听先验整合到CIL设置中，利用其密集表征并引入音频引导视觉的注意力策略，以及特征和logit级别的双重蒸馏目标来缓解遗忘。

Abstract: Class-Incremental Learning (CIL) aims to continuously learn new classes without forgetting previously acquired knowledge. While recent CIL advances have spurred significant interest across various modalities, the audio-visual setting remains underexplored. Furthermore, although foundational multimodal models like SAM-Audio encapsulate rich static priors, our empirical analysis reveals that these representations struggle in incremental settings. This work bridges this gap by integrating SAM-Audio’s audio-visual priors into the CIL setting. Specifically, we leverage its dense audio and visual representations and employ a novel guided attention strategy where the audio features contextually guide the visual representations. To further mitigate catastrophic forgetting, we introduce dual-level distillation objectives at both the feature and logit levels. Extensive evaluations on audio-visual CIL benchmarks demonstrate that our approach consistently outperforms state-of-the-art methods.

[66] Beyond Model Size: Probing the Gaps in Visual in-Context Learning by Training a Tiny Model cs.CVPDF

Sunil Khatri, Steven Landgraf, Markus Ulrich, Simon Reiß

TL;DR: 这篇论文通过训练一个仅有100万参数的微型视觉上下文学习模型，与参数规模大7000倍的大型模型进行对比，揭示了当前视觉上下文学习评估方法在任务编码、预训练任务选择和度量指标方面存在的缺陷。

Details

Motivation: 当前视觉上下文学习方法依赖模型和数据规模的扩展，但作者质疑这是否是实现视觉模型上下文学习能力的关键，因此通过极端反例（微型模型）来检验大型模型的适应能力。

Result: 实验在图像数据分布小偏移、未见过的任务编码和全新任务三种自适应设置下进行，结果表明尽管存在巨大的训练资源差距，当前评估方法未能有效衡量模型的适应能力。

Insight: 创新点在于通过训练微型模型作为基准，暴露了当前视觉上下文学习基准测试的局限性，强调了在自适应能力评估方面需要创新，而不仅仅是扩大模型规模。

Abstract: Visual in-Context Learning (VICL) aims at making progress towards adaptive vision models, that can – based on a few examples – adapt to a new task at test-time. With the history of in-context learning in natural language processing research, where large, parameter-heavy models are in use, one pathway that current VICL methods take is model- and data-scaling as key ingredients. Yet, it is not clear, whether these ingredients are the key for in-context learning to take shape in vision models. To stress-test such large models, we challenge them with an extreme counterexample: we train a tiny visual in-context model with merely $1$ million parameters and a modest amount of $70,000$ images. We compare the results of this severely capacity capped tiny model to $7,000\times$ larger VICL models in different adaptive settings, (1) on image data with small distribution shifts, (2) on unseen task encodings and (3) on a completely new task, i.e., the setting VICL envisions. With the chasm of training resources between the tiny- and large models, our experiments showcase a lack in how adaptive capabilities are measured, with respect to how tasks are encoded, which tasks were used in pre-training and the choice of metrics. These gaps in current VICL benchmarking underscore a need for innovation in evaluation of adaptive capabilities.

[67] Quo Vadis, Visual In-Context Learning? A Unified Benchmark Across Domains and Tasks cs.CVPDF

Pradnya Halady, Jiale Wei, Zdravko Marinov, Alexander Jaus, Simon Reiß

TL;DR: 该论文构建了一个名为VIBE的综合性视觉上下文学习基准测试，涵盖广泛的图像领域和任务类型，旨在更准确地评估视觉上下文学习模型在面对新图像和任务分布时的自适应能力。研究对六个模型在14个数据集和12个任务（共106种组合）上进行了压力测试，并揭示了当前视觉上下文学习的状态、局限性、系统性失败模式及有前景的方向。

Details

Motivation: 当前视觉上下文学习的评估局限于狭窄的设置，主要反映预训练中的任务或图像领域，未能真正测试模型的自适应能力。论文旨在填补这一空白，通过构建一个多样化的基准来更清晰地评估模型在新分布下的表现。

Result: 在统一的、可重复的评估协议下，对六个模型进行了单样本设置的压力测试，涵盖了106种数据集-任务组合。评估结果揭示了当前视觉上下文学习的关键见解，包括其局限性和系统性失败模式。

Insight: 论文的主要创新点是构建了一个统一的、跨领域和任务的视觉上下文学习基准（VIBE），这为系统评估模型的自适应能力提供了标准化工具。从客观角度看，这种基准的建立有助于推动该领域从狭隘评估转向更全面、真实的性能检验，并公开工具包促进了更广泛的评估和比较。

Abstract: Visual in-context learning has been proposed as a pathway towards dynamic models that can generate predictions based on a provided context and thereby can adapt to new vision tasks at test-time. Yet, the evaluation of the adaptation capabilities of these models has been limited to narrow setups that mainly mirror tasks or image domains from pre-training for which real adaptation is not required. We address this gap by constructing a broad Visual In-Context BEnchmark (VIBE) with a focus on diverse imaging domains and a wide range of tasks. With this, we are able to get a much clearer picture of the adaptive capabilities of visual in-context models when faced with new image- and task distributions. We stress test six models on $14$ datasets and $12$ tasks (in total, we explore $106$ dataset-task combinations) and compare them under a unified, reproducible evaluation protocol, in an one-shot setting. Our evaluation uncovers key insights on the state of visual in-context learning, including limitations, systematic failure modes and promising directions. To foster broader evaluation, we will openly release our VIBE toolkit.

[68] The 1st PortraitCraft Challenge: A CVPR 2026 Workshop Competition on Portrait Composition Understanding and Generation cs.CVPDF

Zijie Lou, Youyun Tang, Xiaochao Qu, Haoxiang Li, Ting Liu

TL;DR: 本文介绍了CVPR 2026研讨会竞赛PortraitCraft Challenge的概况，该竞赛专注于人像构图理解与生成，旨在推动人像美学分析和可控图像合成领域的研究。竞赛包含两个互补赛道：结构化人像构图理解和基于结构化描述与构图约束的人像生成，并为此构建并发布了包含约5万张精选真实人像图像的大规模数据集。

Details

Motivation: 现有数据集和任务主要关注全局美学评分，该竞赛旨在通过引入一个统一的评估框架，推动人像构图这一更结构化、更可控方向的研究。

Result: 报告描述了挑战赛的设置、评估协议、数据集构成和最终结果，并对提交方案的技术特点进行了分析，为相关研究提供了一个标准化、可复现的平台。

Insight: 创新点在于提出了一个将构图理解与生成任务统一起来的评估框架，并构建了提供多层次监督的大规模人像构图数据集，有望促进人像美学和可控图像生成领域的进一步发展。

Abstract: This paper presents an overview of the inaugural PortraitCraft Challenge, held as one of the official competitions at CVPR 2026. The challenge focuses on portrait composition understanding and generation, aiming to advance AI research in portrait aesthetics analysis and controllable image synthesis. Unlike existing datasets and tasks that primarily focus on global aesthetic scoring, PortraitCraft introduces a unified evaluation framework comprising two complementary tracks. Track 1 requires models to perform structured portrait composition understanding, and Track 2 requires models to generate portrait images from structured composition descriptions under explicit compositional constraints. To support the challenge, we constructed and publicly released a large-scale portrait composition dataset consisting of approximately 50,000 curated real portrait images, providing multi-level supervision. This report describes the challenge setup, evaluation protocols, dataset composition, and final results, along with an analysis of the technical characteristics of the submitted solutions. The PortraitCraft Challenge provides a standardized and reproducible platform for research on portrait composition understanding and generation, and is expected to foster further progress in the fields of portrait aesthetics and controllable image generation.

[69] AnimaSpark: A Feed-Forward Method for Animating Arbitrary 3D Objects cs.CV | cs.GRPDF

Yiming Zhao, Haoyu Sun, Aoyu Wang

TL;DR: 本文提出了一种名为AnimaSpark的通用3D动画生成方法。该方法通过将绑定骨骼的静态3D模型渲染为多层图像表示，利用视频生成模型和关键点跟踪算法获取2D平面运动，最后将运动提升至3D空间以驱动角色动画。

Details

Motivation: 当前通用的3D动画生成方法在推理速度、运动质量和文本提示遵循方面存在局限，导致动画制作仍依赖繁重的手工劳动。本文旨在解决这一瓶颈，实现高效、高质量的类别无关3D动画生成。

Result: 综合评估表明，该方法在文本-运动对齐、运动质量和计算效率等关键指标上均优于现有的最先进技术。

Insight: 核心创新在于洞察到许多基本3D运动对应的关节变换可以在二维子空间中有效建模，并据此设计了一个从2D视频运动到3D关节变换的蒸馏与提升流程，实现了快速且高质量的通用动画生成。

Abstract: While recent advancements in generative AI have substantially accelerated static 3D model creation workflows, the synthesis of category-agnostic 3D animations remains a significant bottleneck in 3D asset production. Current methods for category-agnostic animation generation exhibit critical limitations in inference speed, motion quality, and adherence to textual prompts, thereby leaving the process dependent on labor-intensive manual artistry. To address these challenges, this paper introduces AnimaSpark, a novel pipeline for category-agnostic 3D animation generation. Our approach is motivated by the key insight that for many fundamental motions in the 3D world, the corresponding joint transformations can often be effectively modeled within a two-dimensional subspace. The pipeline begins by rendering a rigged static 3D model into multi-layered image representations of its mesh and skeleton, which are subsequently fed into a video generation model. We then employ a keypoint tracking algorithm on the generated video to capture the motion of the skeletal joints projected onto the camera’s viewing plane. In the final stage, we distill the planar translations and rotations from these tracked keypoints and lift them from the 2D domain into 3D space to animate the character. Comprehensive evaluations reveal that our method achieves superior performance over existing state-of-the-art techniques across key metrics, including text-motion alignment, quality of motion, and computational efficiency.

[70] U-TTT: Towards Generalizable PET Image Denoising via Test-Time Training cs.CVPDF

Zhiwen Yang, Jiayin Li, Hao Lu, Hui Zhang, Zihua Wang

TL;DR: 本文提出了一种名为U-TTT的新型U形模型，用于正电子发射断层扫描（PET）图像去噪。该模型集成了测试时训练（TTT）层，能够在推理过程中通过自监督动态调整参数，以适应每个测试实例的特性，从而解决现有深度学习模型在分布偏移下性能严重下降的问题。

Details

Motivation: 现有PET图像去噪的深度学习模型在分布偏移（如剂量水平或扫描仪类型变化）下性能严重退化，限制了其稳健的临床部署。其根本原因在于传统固定参数模型在训练后无法适应测试数据的变化。

Result: 大量实验表明，U-TTT在具有挑战性的分布偏移（包括未见过的剂量水平和扫描仪）下，实现了最先进的PET去噪性能，并展现出卓越的泛化能力。

Insight: 论文的创新点在于将测试时训练（TTT）范式集成到U形网络中，实现推理时的动态参数适应。从客观角度看，其提出的双域适应机制（包含空间TTT层和频率TTT层）能全面捕捉3D PET数据的复杂退化，分别处理空间结构退化和全局噪声谱/高频细节恢复，这是一个值得借鉴的架构设计。

Abstract: Existing deep learning models for Positron Emission Tomography (PET) image denoising often suffer from severe performance degradation under distribution shifts, fundamentally restricting their robust clinical deployment. This lack of generalization stems from the conventional paradigm of fixed-parameter models that cannot adapt to variations in test data (e.g., dose levels or scanner types) after training. To overcome this limitation and achieve robust generalization, we introduce U-TTT, a novel U-shaped model that integrates Test-Time Training (TTT) layers to dynamically adjust model parameters during inference through self-supervision, thereby adapting to the specific characteristics of each test instance. Furthermore, to comprehensively capture the complex degradations of 3D PET data, U-TTT features a dual-domain adaptation mechanism comprising a Spatial Test-Time Training (S-TTT) layer and a Frequency Test-Time Training (F-TTT) layer. The S-TTT layer captures and corrects spatial structural degradations, while the F-TTT layer suppresses global noise spectra and restores delicate high-frequency details. Extensive experiments demonstrate that U-TTT achieves state-of-the-art PET denoising performance and exhibits superior generalization under challenging distribution shifts, including both unseen dose levels and unseen scanners. Our code will be available at https://github.com/Yaziwel/U-TTT.

[71] IDEAL: In-DEpth ALignment Makes A Discrete Representation AutoEncoder cs.CVPDF

Yitong Chen, Zijie Diao, Junke Wang, Lingyu Kong, Yixuan Ren

TL;DR: 本文提出IDEAL框架，通过联合对齐浅层和深层视觉基础模型特征，提升离散表示自编码器的重建质量。该方法在ImageNet上实现0.61 rFID，并在自回归图像生成任务中达到1.89 gFID的SOTA性能。

Details

Motivation: 现有基于深度VFM特征的表示自编码器因丢失细粒度视觉细节导致重建质量受限，而浅层特征能补充局部外观和结构信息。

Result: 在ImageNet上获得0.61 rFID，比之前最佳方法提升0.28；自回归图像生成任务中达到1.89 gFID，创下新SOTA记录。

Insight: 创新点在于联合利用浅层与深层VFM特征的互补性，使离散视觉标记同时保持视觉保真度和丰富语义，突破了离散化导致信息丢失的瓶颈。

Abstract: Built on pretrained vision foundation models (VFMs), representation autoencoders (RAEs) have recently emerged as a promising approach for constructing semantically rich latent spaces for image generation. However, their reconstruction quality often remains suboptimal, largely because deep VFM representations do not preserve sufficient fine-grained visual detail. This limitation becomes even more severe after discretization, where missing low-level information is difficult to recover. In fact, we observe that shallow VFM features retain considerably richer local appearance and structural detail, which complements the high-level semantics carried by deep features used in existing RAEs. Motivated by this complementary property, we propose Ideal, an In-depth Alignment framework for discrete representation autoencoding. By jointly aligning quantized tokens with both shallow and deep VFM features, Ideal enables the resulting discrete visual tokens to preserve both visual fidelity and rich semantics. Extensive experiments demonstrate that Ideal yields superior reconstruction performance, achieving 0.61 rFID on ImageNet and outperforming the previous best method by 0.28. When used for autoregressive image generation, Ideal further produces a gFID of 1.89, establishing a new state of the art for autoregressive image generation.

[72] WorldOlympiad: Can Your World Model Survive a Triathlon? cs.CVPDF

Yuke Zhao, Wangbo Zhao, Weijie Wang, Zeyu Zhang, Dakai An

TL;DR: 本文提出了WorldOlympiad基准测试，用于从物理真实性、几何一致性和交互保真度三个维度综合评估基于视频的世界模型。该基准将评估分解为物理、几何和交互三个互补的轨道，并覆盖游戏、机器人和通用真实世界视频三大下游场景，旨在暴露现有模型在遵循物理规则、保持3D结构一致性和维持长时程可控交互方面的缺陷。

Details

Motivation: 现有基准测试多关注视觉质量、语义对齐或短期时序一致性，但未能充分评估生成视频是否遵循物理规则、保持连贯的3D结构以及维持长时程可控交互。为弥补这一空白，需要一个新的、结构化的评估协议来诊断世界模型在这些核心能力上的表现。

Result: 在多个最先进模型上的实验揭示了它们在物理推理、3D一致性和长时程交互方面存在显著差距，凸显了当前生成式世界模型的局限性。

Insight: 创新点在于提出了一个多维度、可解释且可扩展的评估框架，将物理规则、几何重建和交互控制等结构化要求系统地纳入世界模型评估，超越了通用的视频质量评估，为未来模型开发提供了更明确的改进方向。

Abstract: We introduce WorldOlympiad, a benchmark for diagnosing video-based world models across physical faithfulness, geometric consistency, and interaction fidelity. While existing benchmarks often focus on visual quality, semantic alignment, or short-term temporal coherence, they provide limited insight into whether generated videos obey physical rules, preserve coherent 3D structure, and sustain controllable interactions over long horizons. To address this gap, WorldOlympiad decomposes world-model evaluation into three complementary dimensions. The physical track uses object segmentation and MLLM-as-judge to assess whether generated videos follow interpretable rules in mechanics, thermal phenomena, and material properties. The geometry track reconstructs generated videos with Gaussian splatting and evaluates structural consistency, cross-view coherence, and camera-trajectory alignment. The interaction track assesses whether generated rollouts follow complex action prompts and maintain smooth, coherent transitions across consecutive video chunks. WorldOlympiad further covers three major downstream scenarios, including gaming, robotics, and general real-world videos, capturing diverse challenges from interactive control and embodied manipulation to open-domain motion and camera dynamics. Together, these tracks and scenarios form a scalable and interpretable evaluation suite that exposes failure modes beyond generic video quality. Experiments on state-of-the-art models reveal substantial gaps in physical reasoning, 3D consistency, and long-horizon interaction, underscoring the need for more structured evaluation protocols for generative world models.

[73] FADA: Accessible fetal ultrasound interpretation and annotation with a selectively distilled unified vision-language model cs.CV | cs.AIPDF

Mahmood Alzubaidi, Uzair Shah, Raden Muaz, Ines Abbes, Nader Mohammed

TL;DR: 本文提出FADA，一个基于Qwen3.5-VL构建的统一视觉语言模型，旨在解决中低收入国家因专业超声医师短缺而导致的产前超声筛查不足问题。该模型通过单一流程，无需外部标签即可完成临床解读、分类、检测和分割任务，并利用选择性蒸馏技术从多个领域专用基础模型中提取知识。FADA-SKD变体在分割、检测和结构化解读方面表现出色，且能在单张消费级GPU上训练，并部署在智能手机上离线运行，为资源有限环境提供了实用的AI辅助胎儿评估方案。

Details

Motivation: 全球范围内训练有素的超声医师短缺，特别是在中低收入国家，导致超过一半的孕妇无法获得专业的超声检查。现有的深度学习方法通常孤立地处理检测、分割或分类任务，每个任务都需要单独的模型和推理时的专家指定标签，这限制了其可及性和实用性。

Result: 推荐的FADA-SKD变体在分割任务上达到0.8820的平均Dice系数，检测任务上达到0.7671的mAP@0.50，结构化解读合规性为100%。在237张图像的专家验证中，73.5%的解读在临床医生指导下获得满分。模型可在单张消费级GPU上训练，并在搭载Qualcomm Snapdragon 7 Gen 1处理器的智能手机上离线运行完整五阶段流程，耗时约60秒。

Insight: 论文的核心创新在于提出了一个‘解读优先’的统一视觉语言模型管道，将临床解读与注释任务（分割、检测）整合。其‘选择性蒸馏’策略（仅对注释任务进行特征对齐，而解读任务依赖标准微调）在多数评估指标上优于完全蒸馏，实现了性能与效率的平衡。这为在资源受限环境中开发可离线部署、多任务集成的AI辅助医疗工具提供了可行的技术路径。

Abstract: A global shortage of trained sonographers limits prenatal ultrasound screening in low- and middle-income countries, where over half of pregnant women receive no skilled sonography. Current deep learning approaches address detection, segmentation, or classification in isolation, each demanding a separate model and expert-specified labels at inference. We present FADA, a unified vision-language model built on Qwen3.5-VL that performs clinical interpretation, classification, detection, and segmentation through a single interpretation-first pipeline without external labels. FADA distills knowledge from four domain-specific foundation models (FetalCLIP, UltraSAM, USF-MAE, UltraFedFM) via offline pre-computed feature caching. Selective distillation, which applies feature alignment only to annotation tasks while interpretation relies on standard fine-tuning, consistently outperforms full distillation across most evaluation axes. The recommended variant, FADA-SKD, achieves 0.8820 mean Dice for segmentation, 0.7671 mAP@0.50 for detection, and 100% structured interpretation compliance. Expert sonographer validation across 237 images confirms clinically acceptable outputs in both autonomous and human-in-the-loop modes, with 73.5% of interpretations scoring perfectly under clinician guidance. The system is trainable on a single consumer GPU and deployable without cloud connectivity. We validate edge deployment by running the compressed 0.8B model on a commodity smartphone (Qualcomm Snapdragon 7 Gen 1, 12 GB RAM) using llama.cpp with GGUF quantization, completing the full 5-phase pipeline in approximately 60 seconds entirely offline. This establishes a practical pathway for integrating AI-assisted fetal assessment with portable ultrasound devices, directly addressing diagnostic access gaps in resource-constrained settings. Code, models, and data are available at https://github.com/mahmoodphd/FADA.

[74] P3D-Bench: Benchmarking MLLMs for Parametric 3D Generation and Structural Reasoning cs.CVPDF

Yikang Yang, Zhanpeng Hu, Youtian Lin, Mengqi Zhou, Jingxi Xu

TL;DR: 该论文提出了P3D-Bench，一个用于评估多模态大语言模型在参数化3D生成和结构推理方面能力的基准测试。该基准覆盖了文本到3D、图像到3D和装配到3D三大任务族，并评估生成代码的可执行性、几何保真度、拓扑结构、语义对齐和部件级结构。评估发现，装配任务最具挑战性，模型在精确参数化几何和部件级结构恢复方面普遍较弱。

Details

Motivation: 现有基准很少通过代码来评估3D建模能力，而参数化3D建模不仅要求代码可运行，还需要生成的程序在几何上精确、语义上对齐且装配一致。因此，需要一个新的基准来系统评估模型在这些方面的能力。

Result: 在400个文本案例、400个图像案例和203个带标注的装配案例上评估了前沿的MLLM和纯文本LLM。结果表明，装配任务是最困难的设置，模型在精确参数化几何和部件级结构恢复方面表现不佳，尽管它们通常能恢复目标的整体形状和语义身份。

Insight: 论文的创新点在于提出了首个专注于参数化3D程序生成的综合基准，强调了对设计结构（而不仅仅是外观）的恢复能力。其评估协议统一且多维，为未来模型在精确几何和结构化推理方面的改进提供了明确的评估方向。

Abstract: Multimodal large language models can write code to produce complex programs as well as use programs to do 3D modeling, which opens up a new avenue for 3D generation powered by their priors, world knowledge and reasoning. Yet existing benchmarks rarely evaluate 3D modeling through code. Such modeling demands more than runnable code: from a text or visual specification, a model must generate a parametric 3D program that is geometrically precise, semantically aligned and assembly-consistent. We introduce P3D-Bench, a benchmark for parametric 3D generation. Unlike a 3D mesh, a parametric 3D program exposes explicit dimensions, construction operations and part relations, revealing whether a model recovers a design’s structure, not just its appearance. Under a unified protocol, P3D-Bench covers three task families (Text-to-3D, Image-to-3D and Assembly-3D) and scores each output for executability, geometric fidelity, topology, text-grounded constraints, multiview semantic alignment and part-level structure. We evaluate frontier MLLMs and text-only LLMs on 400 text cases, 400 image cases and 203 annotated assemblies, with domain-specific models as reference points. Our extensive evaluation yields three findings. First, assemblies are the hardest setting, where models still fail to compose multiple parts into a coherent structure. Second, models can often recover the global shape and semantic identity of the target object, yet fail to reproduce the precise parametric geometry specified by the input. Third, part-level modeling remains weak on assemblies, where models recover neither the geometry of each part nor the right number of parts. These results position P3D-Bench as a benchmark for evaluating precise parametric geometry and part-level structure in parametric 3D generation.

[75] Data Journalist Agent: Transforming Data into Verifiable Multimodal Stories cs.CV | cs.CL | cs.CY | cs.HCPDF

Kevin Qinghong Lin, Batu EI, Yuhong Shi, Pan Lu, Philip Torr

TL;DR: 本文介绍了Data Journalist Agent (Data2Story)，一个将原始数据转化为可验证的多模态新闻故事的多智能体框架。它通过协调不同专业角色（如数据分析、视觉设计）来模拟虚拟新闻编辑室，旨在实现端到端的自动化新闻报道，并强调证据可追溯性和多媒体内容生成。

Details

Motivation: 解决高质量新闻报道需要新闻团队花费数周时间进行数据分析、角度选择和视觉设计的问题，探索智能体能否端到端地完成数据记者的工作，将原始信息转化为非专家也能信任的故事。

Result: 在18篇文章上与专家原作进行对比评估，结果显示Data2Story在透明度和可审计性方面表现突出，能生成具有竞争力的、证据可追溯的多媒体故事；人类文章在编辑角度、创意设计和呈现方面仍具优势。评估维度包括角度覆盖、人工评分、计算机代理导航和可验证性检查。

Insight: 创新点包括：(i) 证据可追溯性：通过Inspector角色将每个数字、角度和素材链接回数据、代码或外部参考，确保声明有据可依；(ii) 多模态生成：根据读者需求推理并部署交互式地图、音频等多媒体工具，超越纯文本和静态图表。这为记者提供了基于证据、透明且可验证的协作工具。

Abstract: Data tells stories that shape society; the data journalist’s job is to turn raw information into stories non-experts can trust. A high-quality news feature takes a newsroom team weeks: hunting for context, running statistics, choosing an angle, and designing visuals. Recent agents handle individual steps well: data-science agents close the analysis loop, while design agents synthesize beautiful websites. But can an agent serve as a data journalist end to end? We introduce Data Journalist Agent (Data2Story), a multi-agent framework that orchestrates specialized roles into a single virtual newsroom. Data2Story contributes two innovations. (i) Claims are evidence-grounded: an Inspector links every number, angle, and asset back to data, code, or an external reference. (ii) Articles are multimodally generative: rather than defaulting to plain text and static charts, Data2Story reasons about what readers will want to see, then deploys multimodal tools, such as interactive maps for geography and audio for music. We evaluate Data2Story on 18 articles, each paired with the originally published expert piece, along four axes: (a) human-agent angle coverage; (b) rubric evaluation with 53 participants across five dimensions; (c) computer-use agents as judges, a cost-saving proxy for how readers navigate interactive articles; and (d) verifiability, where a coding verifier re-executes statements against the data and checks claims against references. Data2Story produces competitive, evidence-traceable multimedia stories, with particular strength in transparency and auditability. Human articles retain an edge in editorial angle, creative design, and presentation. We position Data2Story as a collaborator for journalists, enabling more evidence-based, transparent, and verifiable reporting. Code and demos are available at https://data2story.github.io.

[76] Lip Forcing: Few-Step Autoregressive Diffusion for Real-time Lip Synchronization cs.CVPDF

Paul Hyunbin Cho, Jinhyuk Jang, SeokYoung Lee, Joungbin Lee, Siyoon Jin

TL;DR: 本文提出Lip Forcing，一种用于视频到视频（V2V）唇形同步的自回归扩散方法。该方法通过将140亿参数的音频条件双向视频扩散模型（教师模型）蒸馏为因果学生模型，实现了实时推理。学生模型仅需两步去噪即可生成每个视频块，无需推理时的分类器自由引导（CFG），从而大幅提升速度。

Details

Motivation: 解决现有基于扩散的唇形同步模型因使用全序列双向注意力机制和大量去噪步骤而无法进行实时推理的问题，旨在实现高质量且实时的唇形同步。

Result: 在两种学生模型规模上验证了方法：13亿参数的学生模型达到31 FPS，比同规模双向模型快17.6倍；140亿参数的学生模型（目前V2V唇形同步中最大的扩散模型）在保持可比参考保真度的同时，比其教师模型快39.8倍。首帧生成时间低于毫秒级，远超所有扩散基线模型。

Insight: 创新点包括：首次将自回归扩散应用于V2V唇形同步；通过教师轨迹分析揭示了CFG在保真度与同步性之间的权衡，并据此设计了三个关键组件（Sync-Window DMD、两步推理调度和基于SyncNet的奖励）；通过蒸馏实现了从大模型到高效实时模型的转化，在速度与质量间取得了良好平衡。

Abstract: Diffusion-based lip synchronization models achieve strong visual quality and audio-visual alignment, but full-sequence bidirectional attention and many denoising steps make them impractical for real-time inference. We present Lip Forcing, to our knowledge the first autoregressive diffusion method for video-to-video (V2V) lip synchronization, which distills a 14B audio-conditioned bidirectional video diffusion teacher into causal students. At inference, the students generate each chunk in only two denoising steps without inference-time CFG, enabling real-time lip synchronization. A lip-sync-specific teacher-trajectory analysis reveals a CFG fidelity-sync tradeoff: no-CFG predictions favor reference fidelity, whereas CFG-guided predictions favor synchronization within a mid-trajectory band. Lip Forcing translates this finding into three analysis-derived components: Sync-Window DMD, a two-step inference schedule, and a SyncNet-based reward. We validate Lip Forcing at two student scales, both distilled from the 14B teacher. The 1.3B student crosses into real-time streaming at 31 FPS, $17.6\times$ faster than its same-scale bidirectional model. The 14B student, the largest diffusion model reported for V2V lip synchronization, runs $39.8\times$ faster than its teacher at comparable reference fidelity. Time-to-first-frame is sub-millisecond at both scales, far below every diffusion baseline.

[77] AnyMod-LLVE: Low-Light Video Enhancement with Modality-Agnostic Inference cs.CVPDF

Hangfeng Liang, Yutao Hu, Yanhan Hu, Xiaohan Wu, Wenqi Shao

TL;DR: 本文提出了一种名为AMNet的统一多模态低光视频增强框架，支持灵活的模态无关推理，即在辅助模态（如事件流或红外图像）可能缺失的情况下仍能进行增强。该方法通过空间-光谱双门控转换器学习辅助模态与RGB输入之间的对应关系，生成隐式辅助表示以支持鲁棒的增强。此外，基于仅RGB数据集与合成辅助模态进行了大规模多模态预训练，以促进跨模态对应学习。实验表明，AMNet能够处理任意的推理时模态组合，并在模态缺失条件下表现出优越的低光视频增强性能。

Details

Motivation: 解决现有多模态低光视频增强方法在推理时通常假设辅助模态可用，而实际场景中这些模态往往不可行的问题，旨在实现模态无关的推理。

Result: 大量实验证明，AMNet能够处理任意推理时模态组合，并在模态缺失条件下表现出优越的低光视频增强性能。

Insight: 创新点在于提出了支持模态无关推理的统一多模态框架，并引入了空间-光谱双门控转换器来学习跨模态对应关系，生成隐式辅助表示；同时，通过基于合成辅助模态的大规模预训练来促进学习，这为实际部署中处理模态不确定性提供了实用方案。

Abstract: Low-light video enhancement (LLVE) remains a challenging task due to severe information degradation under low-illumination conditions. Recent multimodal approaches have significantly improved enhancement performance by incorporating auxiliary modalities, such as event streams and infrared images. However, these methods typically assume the availability of these modalities at inference, which is often not feasible in real-world scenarios. To solve this problem, in this work, we propose AMNet, a unified multimodal framework for LLVE, to support flexible modality-agnostic inference, where auxiliary modalities may be unavailable. To address the issue of modality absence, we introduce a Spatial-Spectral Dual-Gated Translator that learns the correspondence between auxiliary modalities and RGB inputs, producing implicit auxiliary representations to support the robust enhancement. Additionally, to fully facilitate the learning of cross-modal correspondence, we conduct large-scale multimodal pretraining based on the RGB-only dataset with synthetic auxiliary modalities. Extensive experiments demonstrate that AMNet could handle arbitrary inference-time modality combinations and exhibits superior performance for LLVE under modality absence conditions. Code and models are available on the project page.

[78] Next Forcing: Causal World Modeling with Multi-Chunk Prediction cs.CVPDF

Gangwei Xu, Qihang Zhang, Jiaming Zhou, Xing Zhu, Yujun Shen

TL;DR: 本文提出了Next Forcing，一个用于因果世界建模的多块预测框架，旨在解决现有自回归视频生成方法训练收敛慢、精度有限以及推理速度慢的问题。该方法通过引入轻量级辅助模块，同时预测多个未来时间块的视频内容，从而在训练时提供密集的多尺度时序监督，在推理时实现并行预测以加速。

Details

Motivation: 现有自回归世界动作模型在训练时仅监督当前块，缺乏对未来动态的显式信号，导致训练收敛慢且在高帧率下精度受限；同时，迭代式的视频去噪过程也导致推理速度缓慢。

Result: 在RoboTwin基准测试上取得了新的SOTA结果（Clean/Random上分别为94.1%/93.5%），在50fps下相比LingBot-VA在5k训练步数时实现了93.1%的相对提升和2.3倍的收敛加速。在PhyWorld基准测试上也有显著提升，并在通用视频预训练上实现了超过50%的FVD降低。推理时实现了2倍的加速。

Insight: 核心创新在于将大语言模型中的多令牌预测思想引入视频生成，通过构建跨预测深度的因果链，让近未来预测为远未来预测提供信息，从而为骨干模型提供密集的多尺度时序监督。这种多块预测目标既加速了训练收敛和推理，又提升了模型精度，尤其是在高帧率场景下。

Abstract: Autoregressive video generation has emerged as a powerful paradigm for World Action Models (WAMs). However, existing approaches suffer from slow training convergence and limited converged accuracy, particularly at high frame rates, as the training supervision is confined to the current chunk without explicit signals about future dynamics; they also suffer from slow inference due to iterative video denoising. In this paper, we present Next Forcing, a multi-chunk prediction (MCP) framework for causal world modeling that enables faster training, higher accuracy, and accelerated inference. Inspired by multi-token prediction in large language models, Next Forcing introduces an MCP training objective that augments the main model with lightweight auxiliary MCP modules to simultaneously denoise video chunks at multiple future temporal horizons (next$^1$, next$^2$, next$^3$ chunks). These MCP modules form a causal chain across prediction depths, where intermediate features fused from multiple layers of the main model are leveraged to predict future dynamics, allowing near-future predictions to inform farther-future ones and providing dense multi-scale temporal supervision back to the main model. During training, the MCP modules significantly accelerate convergence and improve converged accuracy, especially at high frame rates: at 50 fps, Next Forcing achieves a 93.1% relative improvement over LingBot-VA at 5k training steps and 2.3x faster convergence, and establishes new state-of-the-art results on the RoboTwin benchmark (94.1/93.5% on Clean/Random). At inference, the MCP modules can be retained to predict the next video chunk in parallel with the current one, achieving 2x inference acceleration. Next Forcing also demonstrates significant improvements on PhyWorld, a benchmark evaluating adherence to physical laws in video generation, and over 50% FVD reduction on general video pretraining.

[79] ARM: An AutoRegressive Large Multimodal Model with Unified Discrete Representations cs.CVPDF

Junke Wang, Xiao Wang, Jiacheng Pan, Xuefeng Hu, Feng Li

TL;DR: 本文提出了ARM，一种基于离散表示的自回归大模型，通过统一的离散表示将图像理解、生成和编辑统一在下一个token预测框架中。该模型通过训练一个离散语义视觉分词器将图像映射为紧凑的token序列，然后在大规模文本和图像token序列上训练一个7B的自回归模型，最后应用强化学习来优化任务级目标以提升生成和编辑的对齐性能。

Details

Motivation: 旨在解决多模态任务中理解、生成和编辑任务在统一框架下的整合问题，探索自回归建模与强表示和偏好优化相结合作为可扩展多模态智能基础的可能性。

Result: 在目标任务上性能显著提升，例如WISE整体分数从0.50提高到0.56，GEdit-Bench-EN G_O从5.75提高到6.68，并且强化学习还诱导了文本到图像生成和编辑之间的跨任务协同效应。

Insight: 创新点在于提出了一个统一的离散表示自回归模型框架，将视觉理解、生成和编辑整合；通过多目标监督的离散语义视觉分词器实现共享潜在空间；以及应用强化学习优化任务级目标并发现跨任务协同，展示了自回归建模与表示学习和偏好优化结合的有效性。

Abstract: This paper introduces ARM, a discrete representation-based AutoRegressive Model that unifies image understanding, generation, and editing within a next-token prediction framework. ARM is built on three efforts: first, we train a discrete semantic visual tokenizer that maps images into compact token sequences. Our tokenizer is supervised with multiple objectives that jointly promote semantic discriminability, language alignment and faithful reconstruction, thereby supporting diverse tasks in a shared latent space. With this, we train a 7B autoregressive model over large-scale text and image token sequences, seamlessly developing vision-language perception and generation capabilities. Finally, to further improve preference-aligned behavior for text-to-image generation and instruction-guided editing, ARM applies reinforcement learning (RL) to optimize task-level objectives such as visual quality, instruction adherence, and edit consistency. Surprisingly, the results show that RL not only substantially improves performance on the target tasks (e.g., raising WISE overall from 0.50 to 0.56, GEdit-Bench-EN G_O from 5.75 to 6.68), but also induces cross-task synergy between text-to-image generation and editing. Collectively, these findings highlight autoregressive modeling, when paired with strong representations and preference optimization, as a scalable foundation for multimodal intelligence. Code: https://github.com/wdrink/ARM.

eess.IV [Back]

[80] Multimodal Brain Tumour Classification Using Feature Fusion eess.IV | cs.CV | cs.LGPDF

Wajih ul Islam, Muhammad Yaqoob, Javed Ali Khan, Volker Steuber

TL;DR: 本文提出了一种用于脑肿瘤分类的双分支多模态网络，结合了原始MRI扫描图像和91个提取的影像组学特征（包括强度、纹理、形状和边界描述符）。该网络通过预训练的CNN编码图像流，专用MLP编码影像组学流，并采用拼接、门控或双向跨模态注意力等策略进行特征融合。在包含7,200张图像的平衡数据集上的九次实验运行中，所有多模态配置均优于单模态基线，其中门控融合取得了96.13%的最佳准确率。

Details

Motivation: 临床医生诊断脑肿瘤时，会综合患者症状、病史以及MRI/CT等多模态定量成像数据形成统一判断，而现有深度学习模型通常仅依赖MRI/CT图像，无法复现这种多模态推理过程。

Result: 在平衡的7,200张图像数据集上进行的九次实验表明，所有多模态配置均优于单模态基线，其中门控融合策略取得了96.13%的最佳分类准确率，实现了高性能的脑肿瘤分类。

Insight: 创新点在于将原始医学图像与手工提取的影像组学特征相结合，构建了双分支多模态网络，并探索了多种特征融合策略（如门控融合和双向跨模态注意力），有效模拟了临床医生的多模态推理，提升了分类性能。

Abstract: Clinicians diagnose brain tumors by synthesizing patient symptoms, medical history, and quantitative imaging data from modalities such as MRI and CT scans into a unified clinical judgement. However, most deep learning models rely on MRI/CT images alone, failing to replicate the clinicians multimodal reasoning. We explore a two-branch multimodal network combining raw MRI scans with 91 extracted radiomic features (intensity, texture, shape, and boundary descriptors) to classify brain tumors into glioma, meningioma, pituitary, and no-tumor. A pre-trained CNN backbone encodes the image stream, whereas a dedicated MLP encodes the radiomic stream. Both streams are fused via concatenation, gated, or bidirectional cross-modal attention strategies. Across nine experimental runs on a balanced 7,200 image dataset, all multimodal configurations outperform unimodal baselines with gated fusion achieving the best accuracy of 96.13%.

cs.LG [Back]

[81] SocraticPO: Policy Optimization via Interactive Guidance cs.LG | cs.AI | cs.CLPDF

Zirui Liu, Jie Ouyang, Qi Liu, Xianquan Wang, Jiayu Liu

TL;DR: 本文提出SocraticPO（苏格拉底式策略优化）框架，通过结合自然语言指导和奖励衰减来增强强化学习中的策略优化。该方法在模型推理错误时引入教师模型提供文本指导，但后续正确回答仅获得衰减奖励，从而避免模型过度依赖外部帮助。

Details

Motivation: 传统强化学习通常使用标量奖励（如二元正确性）监督大语言模型的推理，但这类奖励仅提供优化方向，无法解释如何修正错误推理，容易导致捷径学习和脆弱策略。

Result: 在SciKnowEval的大学生科学推理基准测试中，SocraticPO优于强化的RL和自蒸馏基线方法，消融实验表明定向指导和奖励衰减均为必要组件。

Insight: 创新点在于将苏格拉底式对话引入RL rollout过程，通过文本指导增强模型推理能力，同时设计奖励衰减机制防止策略滥用外部帮助；该方法仅修改rollout而保持标准期望奖励目标不变，可灵活集成到现有策略梯度后端（如Reinforce++），且能利用黑盒教师模型而无需访问logits或进行分布匹配。

Abstract: Reinforcement learning (RL) for large language models usually supervises reasoning with scalar outcome rewards, such as binary correctness. Such rewards provide an optimization direction but rarely explain how a model should revise its mistaken reasoning, which can encourage shortcut learning and brittle policies. We propose \textbf{SocraticPO} (Socratic Policy Optimization), a policy-optimization framework that augments RL rollouts with Socratic-style natural-language guidance. During rollout, the student first answers independently; if the answer is incorrect, a teacher diagnoses the attempt and provides concise corrective guidance, after which the student continues under the expanded context. Crucially, this guidance is paired with reward decay: correct answers obtained after teacher intervention only receive decayed rewards, preventing the policy from treating teacher help as a free path to reward. Since SocraticPO only modifies the rollout process while leaving the standard expected-reward objective intact, it can be plugged into existing policy-gradient backends such as Reinforce++. Moreover, because the teacher provides only text-level guidance, SocraticPO can leverage stronger black-box teacher models without requiring access to logits or distribution matching. On undergraduate-level scientific reasoning benchmarks from SciKnowEval, SocraticPO improves over strong RL and self-distillation baselines. Ablations show that both targeted guidance and reward decay are necessary, with reward decay mitigating reliance on assisted correction.

[82] PreAct-Bench: Benchmarking Predictive Monitoring in LLMs cs.LG | cs.AI | cs.CLPDF

Hainiu Xu, Italo Luis da Silva, Jiangnan Ye, Yuhao Wang, Wei Liu

TL;DR: 本文提出了一个名为PreActBench的基准测试，用于评估大型语言模型（LLMs）在预测性监控任务上的能力。该任务旨在仅根据部分动作轨迹，预测其最终是否会导致不道德行为。研究构建了包含1000对道德与不道德轨迹的数据集，并评估了多种LLMs和安全防护模型，发现即使强大模型在此任务上也面临挑战。

Details

Motivation: 现有LLM安全研究主要关注对完整轨迹的回顾性检测，只能在伤害发生后识别。本文旨在解决一个被忽视的前瞻性安全问题：能否在最终不道德行为执行前，仅根据部分轨迹进行预测和干预。

Result: 在PreActBench基准上，使用提出的Prefix Foresight F1指标进行评估。结果显示，人类表现良好，但即使是强大的模型（包括LLMs、安全护栏模型和潜在探测方法）在预测性监控任务上仍面临显著挑战，性能有待提升。

Insight: 论文的核心创新点在于将安全研究的焦点从回顾性检测转向前瞻性预测，并为此构建了专门的基准测试和评估指标。这强调了在LLM安全中发展面向未来的风险推理能力的必要性。

Abstract: Large language models (LLMs) are increasingly deployed as autonomous agents capable of executing multi-step action trajectories toward a given objective. While existing safety research has focused on detecting unethical behavior from complete trajectories, this paradigm is fundamentally retrospective: it identifies harm only after it has already occurred. In this work, we study a critical yet overlooked safety task, which we term Predictive Monitoring: given only a partial action trajectory, can a model infer whether it will culminate in an unethical action before the overt action is executed? To support this task, we present PreActBench, a benchmark of 1,000 paired ethical and unethical action trajectories spanning five domains. We evaluate a range of LLMs, safety guardrail models, and latent probing methods across varying fractions of the action trajectory using our Prefix Foresight F1 metric. Results show that while humans achieve promising performance, predictive monitoring remains challenging even for strong models, highlighting the need for future-oriented risk reasoning in LLM safety.

Anirudh Sekar

TL;DR: 论文提出了RKSC（Reasoning-Aware KV Cache Sharing），一种无需训练的大语言模型推理框架，旨在消除多分支推理流水线中的两种结构冗余。它通过ASKS机制共享前缀KV缓存，利用CGEE机制实现置信度门控的提前退出，并通过RSBCM管理缓存以防止无限增长。在多个模型和基准测试中，RKSC实现了显著的推理加速，同时保持了极低的错误率。

Details

Motivation: 解决多分支大语言模型推理（如思维链、验证）中存在的计算冗余问题，特别是KV缓存的重复计算和验证过程的不必要计算开销，旨在提升推理效率而无需微调或修改模型架构。

Result: 在五个模型家族（7B-10B）、四个基准测试和1000个问题上的评估表明，RKSC相比无KV缓存基线平均加速3.008倍（峰值3.990倍），相比vLLM等效前缀缓存平均提升1.66倍，且由CGEE机制引入的错误率仅为0.37%。

Insight: 创新点在于将KV缓存共享从精确令牌匹配推广到基于语义相似性（ASKS），并结合了两种互补的提前退出机制（CGEE）来动态跳过或提前终止验证计算。其无需训练、轻量级挂钩的设计具有很强的实用性和部署友好性。

Abstract: We introduce RKSC (Reasoning-Aware KV Cache Sharing), a training-free inference framework that eliminates two structural redundancies in multi-branch LLM reasoning pipelines. ASKS (Attention-Similarity KV Sharing) computes the prefix KV cache once and broadcasts it to all semantically similar branches via hidden-state cosine similarity, strictly generalising the token-exact prefix caching used by vLLM and SGLang. CGEE (Confidence-Gated Early Exit) applies two complementary exit mechanisms: (1) it skips the verification forward pass entirely when generation confidence is decisive across branches, and (2) it terminates the verification pass at an intermediate layer when per-layer entropy stabilises, using lightweight hooks on the transformer backbone. RSBCM (Reasoning-Selective Block Cache Manager) prevents unbounded cache growth via attention-weighted depth-priority eviction. Across five model families (7B-10B), four benchmarks, and 1,000 evaluated problems, RKSC achieves a mean speedup of 3.008x over the No-KV baseline (peak 3.990x), a 1.66x mean improvement over vLLM-equivalent prefix caching, with a CGEE-induced error rate of only 0.37% (6 errors out of 1,616 verify calls). No fine-tuning or architecture changes are required. Code is available at https://github.com/AnirudhSekar/RKSC.

[84] ERAlign: Energy-based Representation Alignment of GNNs and LLMs on Text-attributed Graphs cs.LG | cs.AI | cs.CLPDF

Xianlin Zeng, Fan Xia, Xiangyu Chen

TL;DR: 本文提出了一种基于能量的表示对齐框架ERAlign，用于在文本属性图上对齐图神经网络和大型语言模型的表示。该方法通过能量模型将图结构编码和文本嵌入投影到共享潜在空间，实现分布一致性，并在八个数据集上取得了最先进的性能。

Details

Motivation: 现有方法在整合GNN和LLM处理文本属性图时，主要依赖启发式粗粒度匹配，缺乏充分约束且忽略分布对齐，导致表示漂移和泛化能力有限。

Result: 在八个文本属性图数据集上的实验表明，ERAlign在不同监督水平和跨任务迁移场景下均达到了最先进的性能。

Insight: 创新点在于引入基于能量的模型进行层间对齐，并提出能量差异来避免高采样成本，同时保证了训练效率和减少能量景观扭曲的理论保证。

Abstract: Text-attributed Graphs (TAGs) incorporate textual node attributes with graph structures to describe rich relational semantics. Recent efforts to integrate Graph Neural Networks (GNNs) and Large Language Models (LLMs) have shown promise for learning on TAGs, yet achieving well-aligned representations remains challenging. Prior studies largely rely on heuristics that perform coarse-grained matching. They lack sufficient constraints and ignore distributional alignment, leading to representation drift and limited generalization. Building on Energy-based Models (EBMs), we propose an Energy-based Representation Alignment (ERAlign) framework that projects GNN-encoded graph structure and LLM-derived text embeddings in a shared latent space to achieve distribution consistency. Concretely, layer-wise alignment is quantified by a distance metric and optimized via an EBM objective. By decreasing energy values, our framework yields well-aligned representations for downstream tasks. During training, we introduce Energy Discrepancy (ED) to avoid high sampling costs associated with intractable normalization. ED also carries theoretical guarantees of higher training efficiency and reduced energy landscape distortion. Empirical evaluations on eight TAG datasets demonstrate that ERAlign obtains state-of-the-art performance across varying levels of supervision and cross-task transfer scenarios.

[85] Representation-Aware Advantage Estimation: Your Reward Model Provides More Than A Scalar Output cs.LG | cs.CLPDF

Guozheng Li, Xiyan Fu, Yiwen Guo

TL;DR: 本文提出了一种名为表示感知优势估计（Representation-Aware Advantage Estimation）的新方法，旨在改进基于人类反馈的强化学习（RLHF）。该方法的核心是图基优势估计（GraphAE），它利用奖励模型（RM）的隐藏状态作为辅助信号，通过图传播在采样组内计算优势值，从而更有效地利用RM编码的细粒度偏好信息。

Details

Motivation: 当前RLHF方法主要依赖奖励模型输出的标量奖励，但这些标量奖励通常有噪声且难以捕捉细微的偏好差异，而RM的隐藏状态包含了更丰富的语义和偏好信息，因此需要一种方法来利用这些信息以改进优势估计。

Result: 将GraphAE集成到GRPO、GSPO和RLOO等现有基于组的RL算法中，在多个基准测试上取得了一致的提升：在Arena-Hard-v0.1上最高提升+6.3，在AlpacaEval 2.0上提升+8.27，在MT-Bench上提升+0.22，证明了该方法能带来更高效和鲁棒的RLHF。

Insight: 创新点在于将RM的隐藏状态（而非仅标量输出）建模为辅助信号，并通过构建图结构（节点为响应，边为隐藏空间相似性）进行图传播来计算优势值，这是一种轻量级且可无缝集成到现有算法中的表示感知优势估计方法，有效利用了RM的丰富表示信息。

Abstract: Current reinforcement learning from human feedback (RLHF) methods primarily rely on scalar rewards from a trained reward model (RM). While effective, scalar rewards are often noisy and fail to capture fine-grained preference differences, whereas RM hidden states encode richer semantic and preference information. We introduce the representation-aware advantage estimation, which leverages RM hidden states and models them as auxiliary signals for better advantage estimation. Specifically, we propose the Graph-based Advantage Estimation (GraphAE), treat each sampled group as a graph, where nodes correspond to responses and edges capture their similarity in the RM hidden space. Then advantages are computed via graph propagation, enabling each sample to incorporate contextual information from its neighbors. GraphAE is lightweight and can be seamlessly integrated into existing group-based RL algorithms. We apply GraphAE to GRPO, GSPO and RLOO, and conduct extensive experiments on different models and benchmarks. Empirical results show consistent improvements across three benchmarks, with gains of up to + 6.3 on Arena-Hard-v0.1, + 8.27 on AlpacaEval 2.0, and + 0.22 on MT-Bench. These results demonstrate that leveraging RM representations leads to more sample efficient and robust RLHF.

[86] How Does Reasoning Flow? Tracing Attention-Induced Information Flow for Targeted RL in LLMs cs.LG | cs.CLPDF

Zhichen Dong, Yang Li, Yuhan Sun, Weixun Wang, Yijia Luo

TL;DR: 本文提出FlowTracer，一种针对大语言模型（LLM）的强化学习（RL）框架，旨在解决细粒度token级信用分配问题。该方法通过构建基于注意力权重的有向无环图来追踪从问题到答案的信息流，并利用全局流结构来分配token信用，从而引导RL奖励信号聚焦于关键推理步骤。

Details

Motivation: 当前LLM的RL方法在token级信用分配上存在不足，通常平等对待所有token，无法区分关键推理步骤与无关的格式化或填充内容。现有利用模型内部信号的方法多为忽略信息传播全局结构的点式启发式方法。

Result: FlowTracer在一系列推理任务上实现了持续的性能提升，表明其能有效引导学习信号聚焦于对正确答案有贡献（或偏离）的token。

Insight: 创新点在于将注意力权重聚合为有向无环图的边容量，并通过全局流分析（如流守恒、保留到达答案区域的影响）来识别信息流主干和高影响力枢纽token，从而为RL提供更精确的token级奖励信号。这提供了一种从信息传播的全局结构角度进行信用分配的新视角。

Abstract: Token-level credit assignment remains a key obstacle for reinforcement learning (RL) in large language models (LLMs), where RL recipes typically treat all tokens equally, failing to distinguish decisive reasoning steps from routine formatting or fluent filler. Recent attempts leverage model-internal signals to assign finer-grained credit, but these are often point-wise heuristics that ignore the global structure of information propagation. We propose FlowTracer, an RL framework that traces answer-targeted reasoning flow on an attention-induced directed acyclic graph in which nodes correspond to tokens and edge capacities come from aggregated attention weights and derives token credit from this global structure. The edge capacities are reweighted to retain only the influence that can reach the answer region, while enforcing local flow conservation so intermediate tokens neither lose nor gain effective mass due to path length or irrelevant branches. On this graph, FlowTracer extracts an information-flow backbone connecting the question to the answer and scores tokens by flow throughput, revealing high-impact hubs and aggregation checkpoints that mediate long-range dependencies. These derived importances are used to shape token-level rewards, enabling learning signals to focus precisely on the tokens that route information toward (or away from) correct answers and delivering consistent performance gains across a range of reasoning tasks.

[87] N-GRPO: Embedding-Level Neighbor Mixing for Enhanced Policy Optimization cs.LG | cs.CLPDF

Xukun Zhu, Hang Yu, Peng Di, Linchao Zhu

TL;DR: 本文提出N-GRPO方法，通过语义邻居混合机制在嵌入层面动态构建输入表示，以增强大型语言模型在数学推理任务中的探索能力。该方法在GRPO框架中引入，旨在生成多样且语义一致的解路径，避免现有方法在冗余性与语义破坏之间的权衡。

Details

Motivation: 当前大型语言模型在数学推理中的rollout阶段面临两难：token级采样常产生仅用词不同的冗余轨迹，而嵌入级随机噪声方法则易破坏语义一致性。

Result: 在DeepSeek-R1-Distill-Qwen系列模型上的实验表明，N-GRPO在数学推理基准上持续超越强基线，并在分布外任务上展现出鲁棒的泛化能力。

Insight: 核心创新是语义邻居混合机制，它通过混合锚定词嵌入与其最近语义邻居的嵌入来注入多样性，同时严格遵循局部语义流形，从而在保持语义一致性的前提下增强探索。这为策略优化中的探索策略设计提供了新思路。

Abstract: The success of Large Language Models in mathematical reasoning relies heavily on the generation of diverse and valid solution paths during the rollout phase. However, current rollout techniques face a fundamental trade-off: token-level sampling often yields redundant trajectories that differ only in rephrasing, while embedding-level methods utilizing random noise frequently disrupt semantic consistency. To resolve this, we introduce N-GRPO, a novel exploration strategy integrated into the Group Relative Policy Optimization (GRPO) framework. Rather than relying on token-level sampling or native embedding-level noise, our approach leverages Semantic Neighbor Mixing. This mechanism dynamically constructs input representations by mixing the embeddings of an anchor token and its nearest semantic neighbors, thereby injecting diversity while strictly adhering to the local semantic manifold. Experimental evaluations on the DeepSeek-R1-Distill-Qwen models across different sizes show that N-GRPO not only achieves consistent improvements over strong baselines on math reasoning benchmarks but also exhibits robust generalization capabilities on out-of-distribution tasks.

[88] AuRA: Internalizing Audio Understanding into LLMs as LoRA cs.LG | cs.AI | cs.CLPDF

Bo Cheng, Lei Shi, Zhanyu Ma, Yuan Wu, Jun Xu

TL;DR: AuRA是一种将音频理解能力内化到大型语言模型（LLM）中的方法，通过LoRA适配和层间蒸馏技术，将ASR编码器的语音表示知识迁移到LLM中，实现了更紧密的语音-语言联合建模和高效的并行端到端推理。

Details

Motivation: 现有方法（如级联ASR-LLM管道、端到端语音-语言模型或基于桥接/蒸馏的适配）存在转录接口延迟、昂贵的多模态训练或顺序语音-语言耦合等问题，AuRA旨在通过轻量级适配和蒸馏克服这些限制。

Result: 在多个语音-语言基准测试中，AuRA在效果和效率上均一致优于级联系统、语音到LLM适配基线以及大规模语音-语言和多模态模型。

Insight: 创新点在于通过层间蒸馏将ASR编码器的语音表示内部化到LLM的LoRA适配中，实现了轻量级、高效的语音-语言联合建模，同时复用预训练模型，避免了大规模多模态训练。

Abstract: Recent efforts to extend large language models (LLMs) to speech inputs typically rely on cascaded ASR-LLM pipelines, end-to-end speech-language models, or bridge/distillation-based adaptation. While these routes respectively reuse strong pretrained components, enable native speech-language interaction, or offer lightweight adaptation, they often suffer from transcript-interface latency, costly multimodal training, or sequential speech-language coupling. To address these limitations, we present AuRA, a method that distills audio encoding capability into the LLM. Specifically, AuRA feeds the same speech input to an ASR encoder (as a teacher) and a LoRA-adapted LLM (as a student) through a lightweight audio embedding layer, and uses layer-wise distillation to align the student’s hidden states with corresponding teacher representations, thereby internalizing speech representations into lightweight LLM-side adaptations. Compared with cascaded and serial bridge methods, AuRA enables tighter speech-language joint modeling and efficient parallel end-to-end inference, while also reusing pretrained speech and language models rather than requiring large-scale multimodal training. On multiple speech-language benchmarks, AuRA consistently outperforms cascaded systems, speech-to-LLM adaptation baselines, and large-scale speech-language and multimodal models in both effectiveness and efficiency.

[89] TRACE: A Unified Rollout Budget Allocation Framework for Efficient Agentic Reinforcement Learning cs.LG | cs.AI | cs.CLPDF

Heming Zou, Qi Wang, Yun Qu, Yuhang Jiang, Lizhou Cai

TL;DR: 本文提出TRACE框架，用于提升多轮智能体强化学习的效率。该框架通过将ReAct式思维-行动-观察轮次建模为语义节点，在固定采样预算下，将资源动态分配到提示根节点和中间前缀节点，以增强奖励对比度。

Details

Motivation: 现有基于可验证奖励的强化学习在多轮智能体任务中，常因奖励对比度不足（如简单/复杂提示反馈方差低、仅结局奖励对多轮决策评估单一）而受限，且以往方法仅在提示层面分配资源，忽略了同一展开过程中各轮次前缀层面的信息量差异。

Result: 在典型智能体基准测试上，TRACE在同等采样成本下实现了有竞争力的性能和效率提升，例如将Qwen3-14B在多跳问答上的平均准确率较基线提升了2.8个百分点。

Insight: 创新点在于将多轮展开建模为树结构，允许预算分配从提示根节点延伸至轮次前缀节点，并通过可泛化预测器估计条件成功概率来指导分配，从而自适应地丰富仅结局反馈并放大策略更新信号。

Abstract: Reinforcement learning with verifiable rewards (RLVR) is a promising approach for enhancing reasoning and agentic behavior in large language models. However, rollout-intensive policy optimization is often limited by insufficient reward contrast, arising when overly simple or complex prompts generate low-variance feedback and when outcome-only rewards assign the same terminal assessment to every decision in a multi-turn rollout. Past efforts have focused on allocating available rollout resources to promising prompts, yet they only leverage sample informativeness at the prompt level and neglect variation in prefix-level informativeness across turns within the same rollout. This work targets multi-turn agentic RL by modeling each ReAct-style thought-action-observation turn as a semantically distinct node, allowing budget allocation to extend from prompt roots to turn-level prefixes with further continuations, which naturally forms tree-structured rollouts. We introduce Tree Rollout Allocation for Contrastive Exploration (TRACE), a unified rollout allocation framework that enhances reward contrast within a fixed sampling budget. Technically, TRACE allocates rollout budget to both prompt roots and intermediate prefixes that are most likely to yield mixed terminal rewards. A shared generalizable predictor estimates conditional success probability at these anchors from prefix histories to guide this allocation. The resulting adaptive tree structure enriches outcome-only feedback and amplifies the policy-update signal. Empirically, TRACE achieves competitive performance and efficiency gains on typical agentic benchmarks, e.g., improving Qwen3-14B Multi-Hop QA average accuracy by 2.8 points over competitive baselines at equal sampling cost.

[90] A Unifying Lens on Supervised Fine-Tuning Through Target Distribution Design cs.LG | cs.AI | cs.CLPDF

Tong Xie, Yuanhao Ban, Yunqi Hong, Sohyun An, Yihang Chen

TL;DR: 本文提出将监督微调（SFT）重新解释为目标分布设计问题，而非单纯最大化演示轨迹中每个标记的似然。作者引入了Q-target框架，将SFT监督分解为对观测标记的依赖强度和对替代标记的概率分配两个明确选择，并基于此提出了Target-SFT方法。该方法在十个推理数据集-模型设置中均表现优异，验证了基于目标分布设计的有效性。

Details

Motivation: 传统SFT通常最大化演示轨迹中每个标记的似然，但观测标记可能非唯一、有噪声或与模型先验不一致，严格拟合这种独热目标可能不是最优的，尤其是在预训练模型已编码丰富知识先验的情况下。

Result: 提出的Target-SFT方法在十个推理数据集-模型设置中均一致优于基线方法，展示了基于目标分布设计的有效性。

Insight: 创新点在于将SFT重新框架为目标分布设计问题，通过Q-target框架显式分解监督选择，统一了多种现有SFT变体。这揭示了SFT训练更基本的设计原则，并为SFT目标开辟了更广阔的搜索空间。

Abstract: Supervised fine-tuning (SFT) typically maximizes the likelihood of every token in a demonstrated trajectory. However, an observed token can be non-unique, noisy, or misaligned with the model prior. Strictly fitting toward this one-hot target may be suboptimal, especially when the pretrained model encodes a rich knowledge prior. In this work, we reinterpret SFT as target distribution design: instead of studying only the loss objective, we analyze the token-level target that the loss drives the model to match. We introduce the Q-target framework, which decomposes SFT supervision into two explicit choices: (1) how strongly to rely on the observed token, and (2) how to allocate the remaining probability mass over alternatives. This perspective unifies many existing SFT variants as implicit choices of the target distribution Q. Building on this view, we propose Target-SFT which constructs the training objective directly from the desired target distribution. This method consistently outperforms across the ten reasoning dataset-model settings evaluated, showing the effectiveness of this target-based approach. Overall, our formulation reveals a more fundamental design principle for SFT training and opens a broader search space for SFT objectives.

[91] Density Ridge Selective Prediction for LLM and VLM Hallucination Detection under Calibration Label Scarcity cs.LG | cs.AI | cs.CVPDF

Nina I. Shamsi

TL;DR: 本文提出了一种基于密度脊的幻觉检测方法，用于在标定标签稀缺的情况下评估大型语言模型和视觉语言模型的输出可靠性。该方法通过构建隐藏状态生成轨迹的核密度估计，将响应流形恢复为密度脊，并计算测试生成的特征点到最近脊顶点的负欧氏距离作为置信度分数。在七个问答基准测试中，该方法在AUROC指标上显著优于现有无监督和监督方法，且在标签稀缺时性能下降更温和。

Details

Motivation: 现有幻觉检测方法中，无监督采样检测器（如语义熵、EigenScore）无需标签但性能存在瓶颈，而监督探针（如SAPLMA）在标定标签稀缺时性能急剧下降。本文旨在解决在标定标签有限的情况下，如何有效检测LLM和VLM的幻觉问题。

Result: 在七个问答基准（HaluEval-QA、TriviaQA、GSM8K、POPE、ScienceQA、A-OKVQA）上，使用九种文本和视觉LLM，并在标签稀缺协议（n_cal=200查询，N=5生成）下评估。基于密度脊的评分在AUROC上比语义熵、SAR、EigenScore、SAPLMA和对数概率等方法提升5-20个百分点，且在标定标签稀缺时表现出更温和的性能下降。

Insight: 创新点在于将LLM的响应流形建模为隐藏状态生成轨迹的密度脊，提供了一种低维几何骨架来表征随机输出分布。该方法通过几何距离评分，在无需大量标定标签的情况下，实现了更鲁棒和高效的幻觉检测，为标签稀缺场景下的模型可靠性评估提供了新思路。

Abstract: Hallucination detection in large language and vision-language models is increasingly framed as selective prediction, where a detector assigns a confidence score and abstains when confidence is low. Unsupervised sampling detectors (Semantic Entropy, EigenScore) avoid labels but plateau in quality, while supervised probes (SAPLMA) attain stronger in-distribution scores yet degrade sharply when calibration labels are scarce. We recover the response manifold of an LLM as the density ridge of a kernel density estimate built on a six-dimensional kinematic feature map of hidden state generation trajectories. A test generation is scored by the negated Euclidean distance from its projected feature point to the nearest ridge vertex, yielding a low-dimensional geometric skeleton of the stochastic output distribution. We evaluate against Semantic Entropy, SAR, EigenScore, SAPLMA, and log-probability on seven QA benchmarks (HaluEval-QA, TriviaQA, GSM8K, POPE, ScienceQA, A-OKVQA) using nine text and vision LLMs in a deliberately label-scarce protocol ($n_{\text{cal}}{=}200$ queries, $N{=}5$ generations). Our ridge-based score beats on AUROC with 5-20 points gain, while demonstrating tempered degradation under calibration-label scarcity.

[92] Geometry-Aware Reinforcement Learning for 2D Irregular Nesting cs.LG | cs.CVPDF

Auguste Lehuger, Guillaume Henon-Just

TL;DR: 本文提出了一种结合几何感知神经编码器与强化学习的2D不规则排样方法，通过Polygons Transformer（PoT）架构编码连续向量几何并实现跨多边形注意力，利用CORL训练框架优化排样策略，在复杂地理轮廓数据集上验证了其性能。

Details

Motivation: 传统启发式求解器在处理2D不规则排样问题时，缺乏对多边形几何的感知，仅依赖有限几何指导的暴力搜索，本文旨在通过强化学习自动从数据中学习几何先验，以更智能地引导探索。

Result: 实验表明，训练后的智能体在面积利用率上达到了与当前最先进启发式求解器Sparrow相当的水平，证明了强化学习能够有效发现并利用几何感知能力。

Insight: 创新点在于提出了Polygons Transformer（PoT）这一几何感知神经编码器架构，结合CORL强化学习框架，实现了从数据中自动学习几何先验以指导排样优化，为空间任务提供了新的学习范式。

Abstract: Traditional heuristic solvers for the 2D irregular nesting problem share a fundamental limitation: they are blind to polygon geometry, relying on guided brute-force to navigate the continuous placement space with minimal geometrical guidance. In this paper, we argue that Reinforcement Learning is uniquely positioned to overcome this bottleneck. By pairing an optimization policy with a geometry-aware neural encoder, an agent can automatically discover rich geometric priors directly from data, utilizing these learned intuitions to strategically guide exploration. To realize this, we introduce the Polygons Transformer (PoT), a novel architecture that encodes 2D continuous vector geometries while allowing cross-polygons attention. We couple this novel architecture with a Combinatorial Optimization Reinforcement Learning (CORL) training framework to find optimal solutions. To support this paradigm, we release an open-source training dataset derived from complex geographic contours alongside a dedicated evaluation benchmark. Our empirical validation demonstrates that our trained agent achieves area utilization performance highly competitive with Sparrow, the state-of-the-art heuristic solver, proving that reinforcement learning can successfully discover and exploit geometric awareness for precise spatial tasks.

cs.AI [Back]

[93] From Senses to Decisions: The Information Flow of Auditory and Visual Perception in Multimodal LLMs cs.AI | cs.CL | cs.CV | cs.SDPDF

Wish Suharitdamrong, Muhammad Awais, Xiatian Zhu, Sara Atito

TL;DR: 本文研究了音频-视觉大语言模型（AVLLMs）中音频和视觉信号的信息流路径，揭示了在音频-视觉视频输入中，信息遵循类似视觉语言模型的顺序流动，而在多交错音频-视觉项目输入中，信息流转向并行处理。研究还发现，一旦信息传递至LLM，音频-视觉及其他类型的token可被丢弃，从而实现更高效的推理。

Details

Motivation: 尽管多模态大语言模型（MLLMs）在研究和实际应用中日益重要，但其内部音频和视觉信号如何影响最终预测的路径尚不清晰，本文旨在揭示AVLLMs中音频-视觉信息流的内部工作机制。

Result: 研究在Qwen2.5-Omni和Video-SALMONN2 Plus等3B和7B规模的多个模型上验证了发现，表明信息流结构在不同任务和数据集上具有普遍性，且丢弃已传递信息的token对预测影响极小甚至略有提升。

Insight: 创新点在于首次系统描绘了AVLLMs中音频和视觉信息的流动与整合机制，提出了信息流路径随输入配置变化的假设，并为模型可解释性、设计和效率优化提供了理论基础，特别是通过token丢弃策略实现高效推理的通用方法。

Abstract: Multimodal Large Language Models (MLLMs) can listen and see, but how do audio and visual signals actually travel through the network to shape an answer? Despite their growing role in research and real-world applications, the internal pathways through which audio and visual tokens influence the final prediction remain poorly understood. In this study, we examine audio-visual information flow inside Audio-Visual Large Language Models (AVLLMs), tracing how AVLLMs route, utilize, and integrate audio and visual information across two input configurations, audio-visual video and multiple interleaved audio-visual items. We find that for audio-visual video, AVLLMs follow the sequential information flow pathway established for VLMs and VideoLLMs, with audio and visual contribution flowing along this pathway in proportion to the task’s reliance on each modality. In settings with multiple interleaved audio-visual items, this routing shifts to different parallel streams. Furthermore, we demonstrate that audio-visual and other token types can be discarded once their information is transferred to LLM, with minimal impact on the model’s prediction or even slight improvement, generalizing across multiple tasks and datasets, enabling more efficient inference. These findings hold across multiple models and scales, Qwen2.5-Omni and Video-SALMONN2 Plus at 3B and 7B scales, leading to hypotheses on why these flow structures emerge. Together, these results deliver the first coherent picture of how AVLLMs orchestrate sound and sight inside the network and lay the groundwork for the next wave of interpretability, design, and efficiency advances in audio-visual and broader MLLMs.

[94] RealMath-Eval: Why SOTA Judges Struggle with Real Human Reasoning cs.AI | cs.CLPDF

Yiteng Mao, Kenan Xu, Yijia Lyu, Wenhao Li, Jianlong Chen

TL;DR: 论文提出了RealMath-Eval基准，包含224份真实世界的高中数学考试答卷，用于评估大语言模型对真实学生多样化推理过程的评判能力。研究发现，即使是SOTA的LLM评判者在该任务上也表现不佳，与专家评分存在较大误差，并且揭示了LLM在评判合成文本与真实学生文本时存在显著的‘评估鸿沟’。

Details

Motivation: 尽管LLM在解决高中数学题上已接近完美，但其评估真实学生多样化推理过程的能力尚未得到充分检验。本文旨在填补这一空白，探究LLM作为评判者（judge）在处理真实人类推理时的局限性。

Result: 在RealMath-Eval基准上，SOTA LLM评判者表现显著不佳，与专家评分的均方误差高达~~2.96。相比之下，在评估合成LLM生成的解答时，其均方误差仅为~~1.17，准确性和一致性更高，揭示了显著的‘评估鸿沟’。

Insight: 核心创新点在于构建了首个基于真实学生数学答卷的评估基准，并系统揭示了LLM评判者对合成数据与真实数据评估能力的不匹配。研究发现，合成错误在语义空间中呈现‘结构性坍缩’，而人类错误则更加多样化且信息论惊奇度更高，这挑战了当前严重依赖合成数据的LLM评估流程的有效性。

Abstract: While Large Language Models (LLMs) have achieved near-perfect performance in \emph{solving} high-school mathematics, their ability to \emph{evaluate} the diverse reasoning processes of real human students remains under-examined. To bridge this gap, we introduce \textbf{RealMath-Eval}, a rigorously annotated benchmark of 224 real-world exam responses from high schools. Our initial evaluation reveals that even state-of-the-art LLM judges struggle significantly on this task, exhibiting a high Mean Squared Error ($\sim$2.96) against expert human grading. To probe a plausible explanation, we contrast this performance with a control setting where the same judges evaluate synthetic LLM-generated solutions. We identify a stark Evaluation Gap'': judges are considerably more accurate and consistent on synthetic text (MSE $\sim$1.17) but struggle to generalize to authentic student reasoning. Through semantic embedding analysis, we find that synthetic errors suffer from a structural collapse’’ into predictable, low-dimensional linear subspaces, whereas human errors form a more diverse error space. Furthermore, generative probability probes suggest that human reasoning involves significantly higher information-theoretic surprisal, indicating that student reasoning transitions are more out-of-distribution for current models. Finally, we find that surface-level style transfer fails to close this gap. Our findings suggest that current LLM evaluation pipelines relying heavily on synthetic data may not adequately capture the diversity of authentic student mathematical reasoning.

[95] Supervised Fine-tuning with Synthetic Rationale Data Hurts Real-World Disease Prediction cs.AI | cs.CL | cs.LGPDF

Buxin Su, Bingxuan Li, Cheng Qian, Yiwei Wang, Jin Jin

TL;DR: 本文通过大规模实验发现，在临床预测任务中，使用合成推理数据进行监督微调（SFT）反而会损害模型性能，特别是在五年阿尔茨海默病及相关痴呆症（ADRD）预测任务上。研究揭示了叙事合理性与判别优化之间的结构性冲突，并指出相同的推理数据在推理时作为演示能提升性能，但在训练时作为目标则导致性能下降。

Details

Motivation: 研究动机是验证广泛假设——即使用合成推理数据进行监督微调能通过教导模型‘为什么’而不仅仅是‘预测什么’来提升临床预测任务性能，特别是在高风险的ADRD预测场景中。

Result: 在504种配置的大规模控制实验中，基于推理的SFT相对于仅使用标签的微调，持续且显著地损害了预测性能；性能下降在不同模型家族和数据规模上均存在，且使用面向推理的基础模型也无法解决。人类专家标注确认生成的推理在医学上是准确且基于患者特定证据的。

Insight: 创新点在于揭示了合成推理数据在训练与推理阶段作用的矛盾性，并首次系统性地证明了在临床预测任务中，基于推理的监督微调可能因叙事合理性与判别优化的结构性冲突而适得其反，为理解何时及如何有效使用推理监督提供了关键见解。

Abstract: Supervised fine-tuning with synthetic rationale data is widely assumed to improve language model performance on clinical prediction tasks by teaching models not just what to predict but why. We test this assumption on five-year Alzheimer’s disease and related dementias (ADRD) prediction from longitudinal health histories. Across a large-scale controlled experiment of 504 configurations, we find that rationale-based SFT consistently and substantially hurts prediction performance relative to label-only fine-tuning. The degradation persists across model families and data scales, and is not resolved by using a reasoning-oriented base model. Crucially, the failure is not explained by poor rationale quality: human expert annotation confirms that the generated rationales are medically accurate and faithfully grounded in patient-specific evidence, and few-shot experiments show that the same rationales improve performance when used as inference-time demonstrations rather than training targets. We identify the root cause as a structural conflict between narrative plausibility and discriminative optimization. We hope our work paves the path toward a more precise understanding of when and how rationale-based supervision helps and when it does not, guiding the responsible development of language models for high-stakes clinical prediction.

[96] Infini Memory: Maintainable Topic Documents for Long-Term LLM Agent Memory cs.AI | cs.CLPDF

Suozhao Ji, Baodong Wu, Zehao Wang, Lei Xia, Qingping Li

TL;DR: 本文提出Infini Memory，一种可维护的基于文本的持久记忆架构，用于长期LLM智能体。它将智能体记忆组织为主题结构化的文档，通过缓冲和定期整合新观察来更新记忆，并在推理时采用迭代工具调用的检索方式。

Details

Motivation: 解决现有记忆系统将观察存储为孤立记录、摘要或索引片段所导致的证据聚合、事实修订和记忆维护困难的问题，为长期LLM智能体提供能够跟踪变化事实并提供跨会话相关证据的持久记忆。

Result: 在MemoryAgentBench基准测试上，Infini Memory取得了64.7%的综合得分。消融实验表明，主题结构维护和迭代证据检查分别提升了长期记忆使用的不同互补方面。

Insight: 创新点在于将记忆视为主题结构化的文档作为语义单元，结合缓冲整合的维护机制和代理驱动的迭代检索过程。这提供了一种将记忆组织、更新和访问统一为可维护文本文档的系统性方法。

Abstract: Long-term LLM agents need persistent memory that can track changing facts and provide relevant evidence across sessions. Existing memory systems often store observations as isolated records, summaries, or indexed fragments, which makes evidence aggregation, fact revision, and memory maintenance difficult. We propose Infini Memory, a maintainable text-based persistent memory architecture that treats agent memory as topic-structured documents. Each topic document serves as a semantic unit for collecting related evidence, preserving metadata, and revising facts over time. New observations are first staged in a buffer and periodically consolidated into coherent textual contexts. At inference time, an agentic retrieval procedure lets the LLM read memory through iterative tool calls rather than a single retrieval step. On MemoryAgentBench, Infini Memory achieves 64.7% overall score. Ablations show that topic-structured maintenance and iterative evidence inspection improve complementary aspects of long-term memory use.

[97] When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models cs.AI | cs.CL | cs.LGPDF

Sai Kartheek Reddy Kasu, Nils Lukas, Samuele Poppi

TL;DR: 本文提出了一种名为CoT-Output 2x2安全矩阵的追踪级诊断框架，用于揭示多轮推理模型中的隐藏故障模式。该框架通过独立标注每个轮次的内在推理和可见输出，识别出四种操作定义的故障类型，并在信息危害场景下对三个蒸馏推理目标进行了评估。

Details

Motivation: 动机在于多轮推理模型的故障在终端分数评估中难以察觉，模型可能在对话早期就锁定不安全立场，但其最终轮次的拒绝率却与稳健对齐的基线模型无异，因此需要一种方法来暴露这些隐藏的时间动态。

Result: 在信息危害场景下，对三个蒸馏推理目标在五种监督条件下进行了评估，收集了6750个轮次级别的观察数据。分析揭示了两种可复现的漏洞：监督悖论（显式监督线索反而增加了对齐伪装率）和上下文注入故障（模型在内部状态安全的情况下仍锁定不安全的外部输出）。

Insight: 创新点在于提出了一个细粒度的、追踪级别的诊断框架（CoT-Output 2x2矩阵），它能够系统性地识别和分类多轮对话中推理与输出不一致的故障模式，特别是新定义的’上下文注入故障’，这凸显了多轮场景下推理不忠实性的具体表现。该框架和释放的数据集为后续的追踪诊断研究提供了支持。

Abstract: Failures in multi-turn reasoning models are largely invisible to terminal-score evaluation. A model can lock onto an unsafe stance early in a long dialogue, yet its final-turn refusal rate may appear indistinguishable from a robustly aligned baseline. To expose these hidden temporal dynamics, we propose a trace-level diagnostic - the CoT-Output 2x2 safety matrix. This framework labels every turn along two independent axes (internal reasoning and visible output), yielding four operationally defined failure cells: robust alignment, alignment faking, overt jailbreak, and a distinct failure mode we term context-injection failure (where the CoT maintains safe reasoning, but the visible output produces harm, highlighting a multi-turn manifestation of reasoning unfaithfulness). We evaluate three distilled reasoning targets against a fixed attacker across five oversight conditions, collecting 6750 turn-level observations on the Information-Hazard scenario. Our analysis reveals two reproducible vulnerabilities: an oversight paradox where explicit monitoring cues paradoxically increase alignment-faking rates rather than suppress them, and a context-injection failure where models lock onto unsafe external outputs despite safe internal states. We release the full dataset of multi-turn dialogues and CoT traces to support follow-up trace-diagnostic research.

[98] Mind the Gap: Can Frontier LLMs Pass a Standardized Office Proficiency Exam? cs.AI | cs.CLPDF

Tengchao Lv, Dongdong Zhang, Jiayu Ding, Yilin Jia, Yuzhong Zhao

TL;DR: 这篇论文评估了前沿大语言模型在通过中国国家计算机等级考试（NCRE）标准办公软件操作任务方面的能力。研究引入了一个包含200个Word、Excel和PowerPoint综合实操任务的评测集，并采用7,118条机器可评分标准进行百分制打分。实验发现，单轮模型最高得分仅为36.6%，而具备执行反馈、迭代修复和更广泛Office访问权限的更强智能体系统也只能达到68.8%，远低于作为基准的社区参考分数95.5%。

Details

Motivation: 当前LLM智能体在计算机自动化部署中加速，但其处理复杂、专业级生产力软件的能力尚未得到充分测试。作者认为Office自动化是评估文档自动化能力的理想环境，因为它需要长程规划与推理、精确的参数配置以及多应用程序集成。

Result: 在基于NCRE的评测中，7个前沿LLM表现出了明显的局限性。单轮模型的最佳得分率（SR）为36.6%，而更强的智能体系统得分率为68.8%，均远低于作为评分合理性检查的社区参考分数95.5%。

Insight: 论文的创新点在于提出了一个基于标准化专业考试（NCRE）的、可机器评分的综合性基准来量化LLM的办公文档自动化能力。客观来看，该研究揭示了尽管代码生成能力有所进步，但实现可靠、细粒度的Office文档自动化对当前基于代码生成的LLM和智能体系统仍是一个重大挑战，这为评估AI在真实世界复杂任务中的实际应用能力提供了新的视角和严谨的基准。

Abstract: The deployment of Large Language Model (LLM) agents for computer automation is accelerating, yet their ability to navigate complex, professional-grade productivity software is largely untested. We argue that Office automation is an ideal environment for benchmarking document-automation capability, as it requires long-horizon planning and reasoning, precise parameter configuration, and multi-application integration. To quantify this capability, we introduce an evaluation based on China’s National Computer Rank Examination (NCRE), featuring 200 comprehensive practical-operation tasks across Word, Excel, and PowerPoint. Each task is scored on a 100-point rubric scale using 7,118 machine-gradable criteria, and Score Rate (SR) denotes the mean percentage of rubric points earned across these tasks. We benchmark 7 frontier LLMs and observe stark limitations: single-turn models score a maximum of 36.6%. A stronger agentic system with execution feedback, iterative repair, and broader Office automation access reaches 68.8%, but remains below the 95.5% community-reference score used as a scoring sanity check. Ultimately, our experiments demonstrate that despite recent advancements in code generation, achieving reliable fine-grained Office document automation remains a significant challenge for current code-generating LLM and agent systems.

[99] A History-Aware Visually Grounded Critic for Computer Use Agents cs.AI | cs.CL | cs.CVPDF

Jaewoo Lee, Zaid Khan, Archiki Prasad, Justin Chih-Yao Chen, Supriyo Chakraborty

TL;DR: 本文提出了HiViG，一种历史感知且视觉接地的测试时框架，旨在解决计算机使用代理（CUAs）在复杂图形用户界面（GUI）环境中存在的短视决策和缺乏视觉接地的问题。该框架通过一个多模态评论器，将过去的交互抽象为紧凑记录，并结合当前屏幕截图对执行坐标进行视觉验证，以在行动执行前拦截错误。

Details

Motivation: 现有用于计算机使用代理的测试时干预方法（如评论器模型）存在两个关键局限：一是主要关注短视的决策循环（例如忘记早期行动），二是缺乏检测错误行动（例如点击错误的UI元素）所需的视觉接地能力。

Result: 在Web、移动和桌面基准测试中，HiViG始终优于现有的标量和语言评论器，将Qwen3-VL-32B的平均成功率提高了5.8%，将Gemini-3-Flash的平均成功率提高了9.0%，并展现出强大的跨平台泛化能力。消融实验表明，宏行动历史缓解了短视规划，视觉接地评论减少了执行错误。

Insight: 核心创新在于将历史感知（通过宏行动历史总结已完成成就）和视觉接地（通过当前屏幕截图验证原始执行坐标）相结合，构建了一个多模态评论器框架。这种设计能有效处理长视野GUI任务中的规划短视和执行错误问题，对测试时扩展至关重要。

Abstract: Various test-time interventions for Computer Use Agents (CUAs), including critic models, have been developed to improve performance through pre-execution action evaluation in complex Graphical User Interface (GUI) environments. However, existing critics suffer from two key limitations: they (1) focus primarily on short-sighted decision loops (e.g., forgetting earlier actions) and (2) lack the visual grounding needed to detect flawed actions (e.g., clicking wrong UI elements). To address these, we introduce HiViG, a History-aware Visually Grounded test-time framework, built around a multimodal critic trained on real GUI trajectories to abstract past interactions into a compact record and to evaluate actions with visual grounding. At test time, HiViG integrates the critic into the policy decision loop to provide macro-action history, which summarizes the policy’s completed achievements, and visually grounded critique, which verifies raw execution coordinates against the current screenshot to intercept errors before execution. Across web, mobile, and desktop benchmarks, HiViG consistently outperforms existing scalar and verbal critics, improving average success rates over the strongest baseline by 5.8% for Qwen3-VL-32B and 9.0% for Gemini-3-Flash, and demonstrates strong cross-platform generalization. Ablations show that macro-action history mitigates short-sighted planning and visually grounded critique reduces execution errors, with both components being critical for test-time scaling in long-horizon GUI tasks.

[100] Architect-Ant: Editable Automatic Furnishing of Architectural Floor Plans cs.AI | cs.CVPDF

Fedor Rodionov, Aleksandar Cvejic, Michael Birsak, John Femiani, Peter Wonka

TL;DR: 本文提出了Architect-Ant框架，用于自动且可编辑地为建筑平面图布置家具。其核心贡献包括：1）引入了AntPlan-270数据集，这是一个包含270个带有房间级家具边界框标注的专业住宅平面图数据集；2）开发了一个基于微调视觉语言模型的框架，该框架使用一种紧凑的、基于坐标的领域特定语言（DSL）来表示家具布局；3）通过生成捕获建筑约束的程序性推理轨迹来监督模型微调，并应用偏好优化来提升布局质量；4）生成的DSL可被栅格化并用于驱动基于Flux的LoRA渲染器，以生成逼真的蓝图风格图像，同时保持可编辑的符号布局。

Details

Motivation: 解决自动家具布置领域因缺乏真实、专业设计且带有物体级家具标注的平面图数据集而进展受限的问题，旨在为房地产可视化、室内设计和建筑工作流提供基础支持。

Result: 实验表明，Architect-Ant能够生成几何上有效且功能上合理的家具布局，为布置更大规模仅有结构的平面图数据集提供了一条可扩展的路径。

Insight: 主要创新点在于：1）构建了首个专业标注的住宅平面图数据集AntPlan-270；2）提出了一种结合程序性推理轨迹监督和偏好优化的可编辑自动布置框架；3）设计了紧凑的DSL表示，实现了符号布局与逼真渲染的解耦与可编辑性。从客观角度看，将建筑约束显式编码为推理轨迹用于模型监督，以及利用偏好优化细化布局，是提升布局功能合理性的有效方法。

Abstract: Furnished floor plans are fundamental to real estate visualization, interior design, and architectural workflows. However, progress in automatic furniture arrangement has been limited by the lack of real, professionally designed floor-plan datasets with object-level furniture annotations. To address this gap, we introduce AntPlan-270, a curated dataset of 270 architectural floor plans with per-room furniture bounding box annotations across ten residential room categories. Building on this dataset, we present Architect-Ant, an editable automatic furnishing framework powered by a fine-tuned vision-language model. Furniture layouts are represented using a compact, coordinate-based domain-specific language (DSL) that encodes object categories and placements relative to the room geometry. To improve spatial reasoning, we generate procedural reasoning traces that capture architectural constraints such as wall alignment, door and window clearance, circulation, fixture compatibility, and room-specific furniture inventories, and use them to supervise fine-tuning of the model. We then apply preference optimization over candidate object placements to further refine layout quality. The generated DSL can be rasterized into semantic masks and used to condition a Flux-based LoRA renderer, producing realistic blueprint-style furnished floor-plan images while preserving the editable symbolic layout. Experiments on layout furnishing show that Architect-Ant produces geometrically valid and functionally plausible layouts, and suggest a scalable path for furnishing larger structure-only floor-plan datasets.

cs.MA [Back]

[101] Decoupling Thought from Speech: Knowledge-Grounded Counterfactual Reasoning for Resilient Multi-Agent Argumentation cs.MA | cs.AI | cs.CLPDF

Jakub Masłowski, Jarosław A. Chudziak

TL;DR: 本文提出了一种名为知识驱动的反事实推理（KG-CFR）的双层架构，旨在提升多智能体辩论框架在长期、受扰动交互过程中的稳定性。该架构通过将私有的、检索增强的规划缓冲区与公共执行层严格分离，来防止逻辑退化、论点重复和角色漂移。在一个名为‘不确定性下的动态资源分配’的专用1v1v1环境中，经过270次完全因子化危机模拟测试，该系统有效防止了超过95%的受扰动运行中出现关键质量下降，并将整体论点质量从0.694提升至0.822。

Details

Motivation: 现有的多智能体辩论框架主要优化最终输出准确性，而忽略了在持续扰动下长期交互过程的稳定性问题，这会导致逻辑退化、论点重复和角色漂移。本文旨在从结构上防止智能体身份丢失并保持过程保真度。

Result: 在自定义的‘不确定性下的动态资源分配’基准环境中，经过大量随机环境冲击的模拟测试，KG-CFR在超过95%的受扰动运行中防止了法官检测到的关键质量下降（定义为质量偏移Δ ≤ -0.20），并将整体论点质量从0.694显著提升至0.822。

Insight: 核心创新点在于通过架构解耦（将私有规划与公共执行严格分离）来系统性增强在持续压力下的韧性，且不损失质量。此外，论文引入了用于衡量话语分歧和规划-执行对齐的自定义向量指标，为操作稳定性提供了有力证据。消融实验表明，适当的理论（知识）基础与前瞻性规划对于论点质量同等重要。

Abstract: Multi-agent debate frameworks have been shown to improve large language model performance in convergent tasks, but they are currently optimized in a way that heavily favors final output accuracy rather than stability of the process. During long-horizon exchanges reactive systems under sustained perturbations often experience logic degradation, argument repetition, and role drift. To structurally prevent the identity loss and maintain the process fidelity, we introduce Knowledge-Grounded Counterfactual Reasoning (KG-CFR), a dual-stage architecture that enforces a strict separation of concerns between a private, retrieval-augmented planning buffer, and a public execution layer. We assess this system in Dynamic Resource Allocation under Uncertainty (DRAU), a dedicated 1v1v1 environment, introducing diversity as distinct from standard debate settings. Over 270 completely factorial crisis simulation trajectories with stochastic environmental shocks, KG-CFR prevents judge-detected critical post-shock degradation (defined as a quality shift, $Δ\le -0.20$) in more than 95% of perturbed runs, increasing the overall argument quality from 0.694 to 0.822. Our primary contribution is the demonstration of architectural decoupling being an important factor of systemic resilience enhancement under sustained pressure without quality loss. Furthermore, we introduce custom vector metrics for discourse divergence and plan-execution alignment that provide strong, directionally consistent evidence of operational stability. Our ablation experiments suggest that the proper doctrinal grounding can be an equally important factor for argument quality, as the prospective planning. KG-CFR, according to our initial metric evaluations, reduces semantic looping, by preserving the agent’s consistency with the original plan.

cs.RO [Back]

[102] GHOST: Hierarchical Sub-Goal Policies for Generalizing Robot Manipulation cs.RO | cs.CV | cs.LGPDF

Sriram Krishna, Ben Eisner, Haotian Zhan, Ying Yuan, Haoyu Zhen

TL;DR: GHOST是一个用于学习视觉运动操作策略的框架，旨在提升策略在训练分布之外的泛化能力。该框架将控制分解为高层策略和低层控制器：高层策略根据多视角RGB-D观测预测下一个子目标（3D末端执行器位姿分布），低层控制器则执行与具体机器人形态相关的动作。通过一个简单的空间接口将3D目标投影到图像平面并以末端执行器热图表示，从而在基于图像的策略中融入3D目标。

Details

Motivation: 解决现有机器人操作策略泛化能力有限的问题，特别是在面对新物体或任务变体时，通过分层控制结构来提升鲁棒性和适应性。

Result: 在一系列操作任务中，与扁平化的Diffusion Policy相比，GHOST的分层分解方法持续提升了性能和鲁棒性。此外，该框架能够轻松整合人类演示，无需依赖（有噪声的）动作重定向，仅需少量人类演示即可适应新物体和任务变体。

Insight: 创新点在于将控制分层为与机器人形态无关的高层子目标策略和形态相关的低层控制器，并引入空间接口将3D目标投影为图像热图。这允许高层策略从人类视频中学习技能应用和组合，而低层策略仅使用机器人数据训练，从而有效结合不同数据源提升泛化能力。

Abstract: We present GHOST, a framework for learning visuomotor manipulation policies that generalize beyond the training distribution. GHOST factorizes control into (i) a high-level policy that predicts the next sub-goal as a distribution over 3D end-effector poses from multi-view RGB-D observations, and (ii) a low-level goal-conditioned controller that executes embodiment-specific actions. To condition image-based policies on 3D goals, we introduce a simple spatial interface that projects predicted goals into the image plane and represents them as end-effector heatmaps. Across a suite of manipulation tasks, this hierarchical factorization consistently improves performance and robustness compared to a flat Diffusion Policy. Further, we show that this hierarchical interface also makes it easy to incorporate human demonstrations without relying on (noisy) action retargeting. As sub-goals are largely embodiment-agnostic, we train the high-level policy on human video to specify how learned skills should be applied and composed, while keeping the low-level policy trained purely on robot data. This hierarchy enables adaptation to novel objects and task variations using a small number of human demonstrations.

[103] Dexterous Point Policy: Learning Point-based Dexterous Hand Policies from Human Demonstrations cs.RO | cs.CV | cs.LGPDF

Beomjun Kim, Seong Hyeon Park, Seunghoon Sim, Seungjun Moon, Sanghyeok Lee

TL;DR: 本文提出Dexterous Point Policy框架，直接从人类演示视频学习灵巧手操作策略，无需机器人演示数据。其核心是使用统一的3D关键点表示（如手腕和指尖）来桥接人类与机器人之间的具身鸿沟，通过自回归Transformer在关键点上训练策略，并在真实机器人任务上实现了显著性能提升。

Details

Motivation: 解决机器人基础模型在人类视频预训练后部署到真实机器人时存在的显著具身鸿沟问题，以及机器人数据收集成本高昂、耗时（尤其在灵巧操作中）的挑战。

Result: 在真实机器人任务（包括拾放和工具使用）上达到75.0%的成功率，而最先进的VLA基线仅1.0%；方法在未见场景（如多对象环境和新对象类别）中表现出强泛化能力。

Insight: 创新点在于利用统一的3D关键点表示（观察和动作）来对齐人类与机器人行为，实现直接策略迁移；客观分析认为，该方法通过抽象到关键点层面避免了昂贵的机器人演示，为跨具身学习提供了高效解决方案。

Abstract: Robotic foundation models pre-trained on human demonstration videos have shown promise, but a significant embodiment gap remains when the resulting policies are deployed on real robots. A common remedy is to fine-tune these models on robot-specific demonstrations. However, robot data collection can be prohibitively expensive and time-consuming, which is particularly acute in dexterous manipulation, e.g., teleoperating a multi-fingered hand for even a single atomic task can take days. To address this, we introduce Dexterous Point Policy, a framework that learns dexterous manipulation policies directly from human videos and requires no robot demonstrations. Our core insight is that a unified 3D keypoint representation can bridge human and robot embodiments when used for both observations and actions. Specifically, we extract 3D keypoints of task-relevant objects and human hands from raw videos, and train an autoregressive transformer over these keypoints. We observe that at the keypoint level, specifically the wrist and fingertips, human and robot behaviors closely align, enabling direct policy transfer. On a suite of real-robot tasks spanning pick-and-place and tool use, Dexterous Point Policy attains 75.0% success, whereas a state-of-the-art VLA baseline reaches only 1.0%. Furthermore, our method generalizes strongly to unseen scenarios, including multi-object environments and novel object categories.

cs.IR [Back]

[104] $τ$-Rec: A Verifiable Benchmark for Agentic Recommender Systems cs.IR | cs.AI | cs.CLPDF

Bharath Sivaram Narasimhan, Karthik R Narasimhan

TL;DR: 本文提出了一个名为τ-Rec的可验证基准测试，用于评估智能体推荐系统。该基准通过可验证的奖励和揭示标记引导机制取代了主观的LLM-as-a-judge评估，并利用结构化目录谓词和pass^k可靠性指标来系统测试智能体的推理一致性。

Details

Motivation: 随着推荐系统向智能体化、多轮对话界面演进，现有的评估范式（如依赖主观且高成本的“LLM作为评判者”）已难以跟上发展步伐，因此需要一种更客观、可验证的评估方法。

Result: 在五个模型家族（GPT-5.4、Claude Sonnet 4.6、Gemini 2.5 Flash、DeepSeek V4 Flash、Qwen3-32B和GPT-5 mini）的九种配置上评估，结果显示存在显著的可靠性悬崖：即使在pass^1指标下最佳模型仅达到约57%，在pass^4下仅约38%，突显了当前对话智能体部署中的关键差距。

Insight: 创新点在于引入了可验证的奖励机制和揭示标记引导，以客观方式控制任务约束在对话中的呈现，从而提供系统化的可靠性测试；从客观角度看，其pass^k指标为评估智能体在多轮交互中的一致性推理能力提供了新思路。

Abstract: As recommender systems transition toward agentic, multi-turn conversational interfaces, evaluation paradigms have struggled to keep pace. Current benchmarks often rely on “LLM-as-a-judge” evaluations, which introduce subjectivity, high costs and inconsistency. We present $τ$-Rec, a benchmark for agentic recommender systems that replaces subjective evaluation with verifiable rewards and a reveal-tagged elicitation (RTE) mechanism that controls how task constraints surface during dialogue. By testing agents against structured catalog predicates and employing a pass^k reliability metric, $τ$-Rec provides a systematic test for consistent reasoning. Our evaluation of nine configurations across five model families – GPT-5.4, Claude Sonnet 4.6, Gemini 2.5 Flash, DeepSeek V4 Flash, Qwen3-32B and GPT-5 mini – reveals a steep reliability cliff, where even the best model achieves only ~57% at pass^1 and ~38% at pass^4, highlighting a critical gap in current conversational agent deployment. All code and data are publicly available at https://github.com/nbharaths/tau-rec.

hep-ex [Back]

[105] Agentic Hybrid RAG for Evidence-Grounded Muon Collider Analysis hep-ex | cs.AI | cs.CL | cs.IR | physics.ins-detPDF

Ruobing Jiang, Dawei Fu, Cheng Jiang, Tianyi Yang, Zijian Wang

TL;DR: 本文提出了一种面向μ子对撞机研究的证据驱动混合检索增强生成（RAG）框架，该框架结合了稀疏词汇检索与稠密语义检索的混合检索器，以及一个用于查询分解、证据扩展和基于证据的答案生成的智能体推理模块。作者构建了该领域的首个检索增强科学问答基准，涵盖文献语料库、检索和答案生成任务。实验表明，该框架在检索效果、答案质量、证据覆盖率和事实基础性方面均优于现有基线。

Details

Motivation: μ子对撞机研究涉及加速器物理、探测器仪器和高能物理现象学等多个领域，相关证据分散在快速增长的异构科学文献中。随着高能物理（HEP）越来越多地探索智能体辅助分析工作流，高效定位、整合和验证科学证据成为关键能力。现有检索增强生成（RAG）框架在科学问答中前景广阔，但如何在保持检索精度的同时集成智能体推理仍是一个主要挑战。

Result: 在作者构建的μ子对撞机领域首个检索增强科学问答基准上进行了广泛评估。结果表明，混合检索提供了最强的检索骨干，而智能体推理在受控证据扩展和答案合成方面最为有效。基于此原则构建的智能体混合RAG框架在检索有效性、答案质量、证据覆盖率和事实基础性方面持续优于代表性的检索和RAG基线。

Insight: 创新点在于提出了一个结合混合检索（稀疏+稠密）与智能体推理的RAG框架，专门针对科学证据的整合与验证。客观来看，其核心贡献是构建了该领域的首个基准，并系统性地验证了混合检索与智能体推理结合的有效性，为未来基于大规模科学文献的HEP分析智能体奠定了基础。

Abstract: Muon collider research spans accelerator physics, detector instrumentation, and high-energy phenomenology, with relevant evidence scattered across a rapidly expanding and heterogeneous body of scientific literature. As high-energy physics (HEP) increasingly explores agent-assisted analysis workflows, efficiently locating, integrating, and verifying scientific evidence becomes an essential capability. While retrieval-augmented generation (RAG) offers a promising framework for scientific question answering, integrating agentic reasoning without compromising retrieval precision remains a key challenge. In this work, we present agentic hybrid RAG, an evidence-grounded RAG framework for muon collider research. The framework combines a hybrid retriever, integrating sparse lexical and dense semantic retrieval, with an agentic reasoning module for query decomposition, evidence expansion, and grounded answer generation. To enable systematic evaluation, we construct the first benchmark for retrieval-augmented scientific question answering in the muon collider domain, comprising a curated literature corpus together with dedicated retrieval and answer-generation benchmarks covering major detector and physics research topics. Extensive evaluation shows that hybrid retrieval provides the strongest retrieval backbone, while agentic reasoning is most effective for controlled evidence expansion and answer synthesis. Built on this principle, agentic hybrid RAG consistently outperforms representative retrieval and RAG baselines in retrieval effectiveness, answer quality, evidence coverage, and factual grounding. Together, the benchmark and framework provide a foundation for evidence-grounded scientific question answering and future HEP analysis agents operating over large-scale scientific literature.

cs.HC [Back]

[106] CANVAS: Captioning Art with Narrative Visual-Audio AI Systems cs.HC | cs.AI | cs.CLPDF

Vignesh Nagarajan

TL;DR: 本文提出了一个名为CANVAS的自动化系统，利用大语言模型和文本转语音技术，为视觉艺术作品生成多感官描述和同步音频旁白，旨在提升盲人和低视力人群对艺术的可访问性。该系统通过Zapier编排，无需人工干预即可将上传的图像转化为丰富的叙事性描述，实现快速、可扩展的无障碍媒体生产。

Details

Motivation: 解决当前替代文本（alt-text）过于简短或缺失，无法传达艺术品的感官、空间或情感特质，导致盲人和低视力人群难以访问视觉艺术的问题。

Result: 在50件艺术品上的定量评估表明，AI生成的描述在词汇多样性、形容词密度和叙事细节上显著优于基线描述，同时保持了可比的易读性水平。统计测试证实了其在丰富性和长度上的显著差异，且整个流程每张图像的处理时间低于20秒，成本低于0.05美元。

Insight: 创新点在于构建了一个全自动、低成本、高效率的多模态（视觉-文本-音频）艺术描述生成工作流，证明了自动化描述在弥合博物馆和数字馆藏可访问性鸿沟方面的潜力，为更广泛的公众参与提供了新思路。

Abstract: Visual art remains largely inaccessible to blind and low-vision (BLV) audiences due to brief or absent alt-text, which rarely conveys the sensory, spatial, or emotional qualities of an artwork. This study presents an automated workflow that generates multi-sensory art descriptions and synchronized audio narration using large language models and text-to-speech services. The system, orchestrated through Zapier, converts uploaded images into rich narrative captions without human intervention, enabling rapid, scalable production of accessible media. Quantitative evaluation across 50 artworks shows that AI-generated descriptions contain significantly higher lexical diversity, adjective density, and narrative detail than baseline captions, while maintaining comparable readability levels. Statistical tests (t-tests, ANOVA) confirm meaningful differences in richness and length, and the full pipeline produces text-plus-audio outputs in under 20 seconds per image at a cost below $0.05. Findings demonstrate that automated captioning can bridge gaps in museum and digital-collection accessibility, with implications for broader public engagement. Future work can incorporate user studies with BLV participants to assess comprehension, preference, and optimal levels of interpretive language.

[107] Sketch-to-Layout: A Human-Centric Computational Agent for Constraint-Aware Synthesis of Modular Photobioreactors cs.HC | cs.CVPDF

Xiujin Liu, Shuqi Li, Yuxin Lin

TL;DR: 本文提出了一种面向模块化光生物反应器（PBR）立面系统的计算框架，包含一个从草图到布局的智能体和一个藻类健康监测管道。该智能体将布局合成建模为约束满足问题（CSP），利用CP-SAT引擎，将稀疏的用户草图作为软先验，并强制执行端口对齐和全局连通性等硬约束，使非专家能够近乎实时地合成可用于制造的配置。同时，通过结合CNN-注意力主干网络和时间排序损失的弱监督方法，系统仅从照片即可量化生物活力，无需绝对真实标签。

Details

Motivation: 建筑集成光生物反应器（PBRs）是实现碳中和建筑的途径，但其部署受到配置复杂性和生物维护的阻碍。本文旨在通过一个计算框架来协调设计意图与物理有效性，降低部署门槛并实现自主维护。

Result: 实验表明，CSP求解器在高达15x15的网格规模上实现了95.5%的成功率。定性评估证实该框架在确保操作完整性的同时保留了设计语义。长期测试显示，视觉模块产生的健康轨迹与14天的生物周期相符。

Insight: 主要创新点在于将模块化PBR布局合成形式化为CSP问题，并利用CP-SAT引擎高效处理软硬约束，实现了人机交互的快速配置生成。另一个创新点是提出了一种无需绝对真实标签的弱监督藻类健康监测方法，结合了CNN-注意力架构和时间排序损失，为低成本、可扩展的碳捕获系统集成交互式合成与计算机视觉提供了可行路径。

Abstract: Building-integrated photobioreactors (PBRs) offer a pathway for carbon-neutral architecture, yet deployment is hindered by configuration complexity and biological maintenance. This paper presents a modular PBR facade system powered by a computational framework reconciling design intent with physical validity. We introduce ‘carbon-neutralization bricks’ featuring integrated vessel-and-conduit geometry; monolithic fluid channels enable ‘plug-and-play’ assembly. To navigate the combinatorial complexity of 14 modular geometries, we develop a Computational Sketch-to-Layout Agent that formulates layout synthesis as a Constraint Satisfaction Problem (CSP). Using the CP-SAT engine, the agent treats sparse user sketches as soft priors while enforcing hard constraints like port alignment and global connectivity. This allows non-experts to synthesize fabrication-ready configurations in near real-time. Furthermore, to facilitate autonomous maintenance, we propose a weakly supervised algae health monitoring pipeline. By employing a hybrid CNN-attention backbone and a temporal ranking loss, the system quantifies biological vitality from photographs without absolute ground-truth labels. Experiments demonstrate the CSP solver achieves a 95.5% success rate on grid scales up to 15 x 15. Qualitative evaluations confirm the framework preserves design semantics while ensuring operational integrity. Long-term tests show the vision module produces health trajectories aligned with 14-day biological cycles, suggesting that integrating interactive synthesis with low-cost computer vision can democratize scalable carbon capture systems.

cs.MM [Back]

[108] MinhwaNet: Faithful but Insufficient Object Grounding in Korean Folk Painting cs.MM | cs.CV | cs.LGPDF

Joonhyung Bae

TL;DR: 本文研究了韩国民俗画（minhwa）中吉祥符号的识别与绘画流派预测之间的关系，发现仅基于符号列表的模型预测流派效果远差于融合图像与策展文本的模型，且强制流派表示基于对象会损害准确性。研究提出了一种‘忠实但不足’的分离配置，通过部分级检测器生成的空间忠实对象证据图，揭示了流派预测依赖于符号的排列方式而非符号本身。

Details

Motivation: 韩国民俗画由少量吉祥符号构成，这些符号在不同绘画流派中反复出现，因此提出一种计算方法：通过识别画中的符号来推断流派。但基于公共语料库（包含完整绘画、双语策展标题和专家对象裁剪）的研究发现，这种方法并不有效，需要探索更有效的多模态方法。

Result: 实验表明，仅基于符号列表的模型在流派预测上表现较差，而融合图像与策展文本的模型更优；强制对象接地会降低准确性。部分级检测器生成的对象证据图在空间上忠实于策展人标注的符号位置，并与基于补丁的代理梯度显著性一致。

Insight: 创新点在于提出了‘忠实但不足’的分离概念，强调流派预测的关键在于符号排列而非符号存在本身；通过部分级解释和证据图，实现了可解释的局部化视觉证据分析，并区分了可迁移的内容标签（如流派）与不可迁移的风格标签（如时代），为长尾文化遗产集合的评估提供了新视角。

Abstract: Korean folk painting (minhwa) is built from a small vocabulary of auspicious symbols, a tiger for protection, a pair of birds for marital harmony, a peony for wealth, that recur across many of its painted genres. This suggests an obvious computational approach, identify which symbols appear in a painting and read the genre from the inventory. Working with a public corpus that pairs whole paintings, eight-field bilingual curatorial captions, and a separate set of expert object crops, we find that this approach does not work. A model given only a list of which symbols a painting contains predicts the genre far worse than a model that fuses the image with the curatorial text, and forcing the genre representation to be object-grounded actively hurts accuracy. The visual evidence on which the genre prediction rests is nonetheless localized and inspectable. A leakage-safe object evidence map projected from a part-level detector is spatially faithful to where curators isolated symbolic objects and to a patch-based surrogate’s own gradient saliency. We name this configuration a faithful-but-insufficient dissociation. The part-level explanation is honest about what the part-level model sees, yet the genre target turns on how symbols are arranged rather than on which ones appear. The same lens separates a content label that survives transfer to held-out source institutions, genre, from a style label that does not, era, a prediction we confirm on two further labels in the corpus. We release the multimodal system, a worked-example reading of one painting’s evidence map against its catalogue, and a set of evaluation cautions that recur in long-tailed heritage collections.

Table of Contents

cs.CL [Back]

[1] Using Probabilistic Programs to Train Inductive Reasoning in Large Language Models cs.CL | cs.AI | cs.LG | stat.MLPDF

[2] BenSyc: Benchmarking Conversational Sycophancy and Human Alignment in LLMs for Bengali Contexts cs.CLPDF

[3] CodeAlchemy: Synthetic Code Rewriting at Scale cs.CL | cs.LGPDF

[4] Pareto-Guided Teacher Alignment for Fair Personalized Text Generation cs.CL | cs.AI | cs.CY | cs.LGPDF

[5] OpenRTLSet: A Fully Open-Source Dataset for Large Language Model-based Verilog Module Design cs.CLPDF

[6] The Confident Liar: Diagnosing Multi-Agent Debate with Log-Probabilities and LLM-as-Judge cs.CL | cs.AIPDF

[7] Early-Token Confidence Predicts Reasoning Quality in Multi-Agent LLM Debate cs.CLPDF

[8] TabClaw: An Interactive and Self-Evolving Agent for Spreadsheet Manipulation and Table Reasoning cs.CLPDF

[9] PADD: Path-Aligned Decompression Distillation for Non-Router Teacher to Guide MoE Student Learning cs.CL | cs.LGPDF

[10] Do Vision-Language Models See or Guess? Measuring and Reducing Textual-Prior Reliance with a Phrasing-Controlled Benchmark cs.CL | cs.CVPDF

[11] LakeQA: An Exploratory QA Benchmark over a Million-Scale Data Lake cs.CL | cs.AIPDF

[12] KCSAT-ML: Probing Reasoning Models with Nationwide-Cohort Human Difficulty cs.CLPDF

[13] WebChallenger: A Reliable and Efficient Generalist Web Agent cs.CLPDF

[14] ParaBridge: Bridging Paralinguistic Perception and Dialogue Behavior in Speech Language Models cs.CL | cs.SD | eess.ASPDF

[15] REAL: A Reasoning-Enhanced Graph Framework for Long-Term Memory Management of LLMs cs.CLPDF

[16] Dep-LLM: Training-Free Depression Diagnosis via Evidence-Guided Structured Multi-factor with Reliable LLM Reasoning cs.CL | cs.AIPDF

[17] Beyond APIs: Probing the Limits of MLLMs in Physical Tool Use cs.CL | cs.AI | cs.CVPDF

[18] It Takes One to Bias Them All: Breaking Bad with One-Shot GRPO cs.CLPDF

[19] Trace Only What You Need: Structure-Aware On-Demand Hypergraph Memory for Long-Document Question Answering cs.CLPDF

[20] Pushing the Limits of LLM Tool Calling via Experiential Knowledge Integration and Activation cs.CLPDF

[21] Does Reasoning Preserve Alignment? On the Trustworthiness of Large Reasoning Models cs.CLPDF

[22] Attention Amnesia in Hybrid LLMs: When CoT Fine-Tuning Breaks Long-Range Recall, and How to Fix It cs.CLPDF

[23] T1-Bench: Benchmarking Multi-Scenario Agents in Real-World Domains cs.CL | cs.AIPDF

[24] Modeling Complex Behaviors: Multi-Personality Composition and Dynamic Switching in Vision-Language Models cs.CL | cs.AIPDF

[25] Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models cs.CL | eess.ASPDF

cs.CV [Back]

[26] SD-GRPO: Verifiable Segment Decomposition for Long-Form Vision-Language Generation cs.CV | cs.AI | cs.LGPDF

[27] WHU-Infra3D: A Full-stack Multi-modal Dataset and Benchmark for 3D Roadside Infrastructure Inventory cs.CV | cs.LGPDF

[28] A Controlled Audit of Pretraining Contamination in Public Medical Vision-Language Benchmarks cs.CV | cs.AI | cs.LGPDF

[29] Interpretable Temporal Facial-Region Motion Analysis for In-the-Wild Parkinson’s Disease Video Classification cs.CVPDF

[30] BiWM: Advancing Open-Source Interactive Video World Models with Bidirectional Autoregression cs.CV | cs.AIPDF

[31] DB-3DME: From Dataset to Benchmark for Human-aligned Automatic 3D Mesh Evaluation cs.CVPDF

[32] FlexPath: Learned Semantic Path Priors for Image-Based Planning cs.CVPDF

[33] iSAGE: A Human-in-the-Loop Framework for Remote Sensing Semantic Segmentation via Sparse Point Supervision cs.CVPDF

[34] A Large Scale Open-Source Image and Video Dataset for Robust Wildfire Detection and Classification cs.CVPDF

[35] Making Time Editable in Video Diffusion Transformers cs.CV | cs.AI | cs.MMPDF

[36] Content-Induced Spatial-Spectral Aggregation Network for Change Detection in Remote Sensing Images cs.CV | cs.AIPDF

[37] CoCoSI: Collaborative Cognitive Map Construction for Spatial Intelligence cs.CVPDF

[38] Vision-Assisted Foundation Model for Solving Multi-Task Vehicle Routing Problems cs.CV | cs.AIPDF

[39] Geometric Coastline Localization using Vision-Language Models cs.CVPDF

[40] 5% > 100%: Flatness Preference is All You Need for Multimodal Parameter-Efficient Fine-Tuning cs.CVPDF

[41] LAFP: Preserving Latent Action Structure in Latent Policy Learning via Flow Matching cs.CVPDF

[42] 3D-CoS: A New 3D Reconstruction Paradigm Based on VLM Code Synthesis cs.CVPDF

[43] GUI-AC: Enhancing Continual Learning in GUI Agents cs.CVPDF

[44] Audio-Visual Exchange-Aware Token Pruning for Efficient Audio-Visual Captioning cs.CVPDF

[45] GRAR: Glass-induced Reflection Artifact Removal in LiDAR Point Clouds cs.CVPDF

[46] PrismAvatar: Pseudo-Multiview Reconstruction and Subpixel Prism Rendering for Real-Time Stereoscopic Communication cs.CV | cs.GRPDF

[47] Improving Adversarial Transferability on Vision-Language Pre-training Models via Surrogate-Specific Bias Correction cs.CV | cs.AI | cs.CRPDF

[48] Segment and Select: Vision-Language Segmentation in 3D Scenarios cs.CVPDF

[49] GaussTrace: Provenance Analysis of 3D Gaussian Splatting Models with Evidence-based LLM Reasoning cs.CVPDF

[50] Globally Localizing Lunar Rover in Pixels via Graph Alignment cs.CVPDF

[51] Can Image Models Imagine Time? ImageTime: A Novel Benchmark for Probing Visual World Modeling Through Spatiotemporal Consistency cs.CV | cs.AIPDF

[52] ManiSplat: Manipulation Trajectory Synthesis from Monocular Video via Decoupled 3D Gaussian Splatting cs.CVPDF

[53] Envision4D: Envisioning Visual Futures via Feed-forward 4D Gaussian Splatting for Autonomous Driving cs.CVPDF

[54] Kwai Keye-VL-2.0 Technical Report cs.CVPDF

[55] Analyzing Training-Free Corruption Detection for Object Detection Datasets cs.CV | cs.DBPDF

[56] FadeMem: Distance-Aware Memory Consolidation for Autoregressive Video Diffusion cs.CVPDF

[57] Vector Map as Language: Toward Unified Remote Sensing Vector Mapping cs.CVPDF

[58] Patient-Level Diagnosis of Acute Myeloid Leukemia via Deep Learning Analysis of Bone Marrow Smear cs.CV | physics.med-phPDF

[59] A Multimodal RGB and Events Dataset for Hand Detection in First-Person View cs.CVPDF

[60] Spatially Selective Self-Training for Unsupervised Building Change Detection cs.CVPDF

[61] HarmoView: Harmonizing Multi-View Constraints for Identity-Consistent Video Generation cs.CVPDF

[62] Earth-OneVision: Extending Remote Sensing Multimodal Large Language Models to More Sensor Modalities and Tasks cs.CV | cs.AIPDF

[63] SCAIL-2: Unifying Controlled Character Animation with End-to-end In-Context Conditioning cs.CVPDF

[64] LIBERO-Occ: Evaluating and Improving Vision-Language-Action Models under Scene-Induced Occlusion via Viewpoint Imagination cs.CV | cs.AIPDF

[65] Listen, Look, and Learn: Learning Without Forgetting through SAM-Audio cs.CVPDF

[66] Beyond Model Size: Probing the Gaps in Visual in-Context Learning by Training a Tiny Model cs.CVPDF

[67] Quo Vadis, Visual In-Context Learning? A Unified Benchmark Across Domains and Tasks cs.CVPDF

[68] The 1st PortraitCraft Challenge: A CVPR 2026 Workshop Competition on Portrait Composition Understanding and Generation cs.CVPDF

[69] AnimaSpark: A Feed-Forward Method for Animating Arbitrary 3D Objects cs.CV | cs.GRPDF

[70] U-TTT: Towards Generalizable PET Image Denoising via Test-Time Training cs.CVPDF

[71] IDEAL: In-DEpth ALignment Makes A Discrete Representation AutoEncoder cs.CVPDF

[72] WorldOlympiad: Can Your World Model Survive a Triathlon? cs.CVPDF

[73] FADA: Accessible fetal ultrasound interpretation and annotation with a selectively distilled unified vision-language model cs.CV | cs.AIPDF

[74] P3D-Bench: Benchmarking MLLMs for Parametric 3D Generation and Structural Reasoning cs.CVPDF

[75] Data Journalist Agent: Transforming Data into Verifiable Multimodal Stories cs.CV | cs.CL | cs.CY | cs.HCPDF

[76] Lip Forcing: Few-Step Autoregressive Diffusion for Real-time Lip Synchronization cs.CVPDF

[77] AnyMod-LLVE: Low-Light Video Enhancement with Modality-Agnostic Inference cs.CVPDF