论文

多维评估-验证奖励(EVR)将评估分解为独立视觉准则；针对每个准则，MLLM Evaluator生成多个候选假设，Verifier在具体视觉证据中grounding每个claim以接受或拒绝，产生可靠且细粒度的奖励信号。A Multi-dimensional Evaluation-Verification Reward (EVR) decomposes evaluation into distinct visual criteria; for each criterion, an MLLM Evaluator generates multiple candidate hypotheses, and a Verifier grounds each claim in concrete visual evidence to accept or reject it, producing reliable and fine-grained reward signals.

提出QWorld，用分位数-分位数匹配目标替代EP，直接将投影后的潜在样本与秩匹配的高斯分位数对齐，从而在尾部保持有效的修正梯度。QWorld is proposed, which replaces EP with a quantile-quantile matching objective that directly aligns projected latent samples with rank-matched Gaussian quantiles, thereby maintaining effective corrective gradients in the tails.

提出Criterion-Distilled Policy Optimization (CriPO)，通过on-policy自蒸馏增强基于rubric的RL，并采用反事实自教师定位负优势rollout中与准则相关的token，将其token级优势翻转为正值，保留本将被抑制的有用模式。Criterion-Distilled Policy Optimization (CriPO) is proposed, which enhances rubric-based RL via on-policy self-distillation and employs a counterfactual self-teacher to locate criterion-relevant tokens in negative-advantage rollouts and flips their token-level advantages to positive values, preserving useful patterns that would otherwise be suppressed.

提出SA，一种Stable Advantage Fusion框架，避免熵崩塌，在所有六个模型-领域设定上持续优于固定系数的GRPO+OPD融合，聚合得分提升0.70%，训练更稳定。SA is proposed, a Stable Advantage Fusion framework that avoids entropy collapse and consistently outperforms fixed-coefficient GRPO+OPD fusion, improving the aggregate score by 0.70% across all six model-domain settings while achieving more stable training.

本文为RGB-NIR低光成像提供新视角：通过引入3D感知神经建模，无需采集干净RGB数据，并能跨不同噪声水平泛化。This paper offers a new perspective for RGB-NIR low-light imaging by incorporating 3D-aware neural modeling, which obviates the requirement for clean RGB data collection, and generalizes across different noise levels.

研究发现privileged shifts无法给出可靠的答案对齐方向，其幅度主要反映反事实敏感性而非token级学习价值；提出Counterfactual Sensitivity Credit Reallocation (CSCR)，作为GRPO的简单扩展，降低高敏感token的credit并对token级优势重新归一化，同时保留原始credit预算与verifier确定的方向。These findings show that privileged shifts fail to provide reliable answer-aligned directions, while their magnitudes primarily reflect counterfactual sensitivity rather than token-level learning value, and propose Counterfactual Sensitivity Credit Reallocation (CSCR), a simple extension of GRPO that reduces credit for highly sensitive tokens and renormalizes token-level advantages to preserve both the original credit budget and verifier-determined direction.

世界模型为规划和行动提供了预测性基础，但现有建模方式仅回答物理层面的问题：它是什么/在哪里，以及将如何演变。然而，人类行为由隐藏的心理状态驱动（一个人相信什么、想要什么、意图做什么、感受如何，以及认为在社会上何为可接受），因此仅追踪物理场景而忽略每个智能体所知与所信内容的模型，会对看起来正确的场景预测出错误的行动。我们将心理世界建模（MWM）形式化为一个通用理论框架，将心理变量作为世界模型的核心组成部分。World models enable a predictive substrate for planning and action, yet existing formulations merely answer a physical question: what/where it is, and how will it evolve. Human behavior, however, is driven by hidden mental state (what a person believes, wants, intends, feels, and considers socially permissible), so a model that tracks the physical scene but not what each agent knows and believes about it predicts the wrong action for the right-looking scene. We formulate Mental World Modeling (MWM), a generic theoretical framework that makes mental variables core components of a world model ra

提出EMBL AI Librarian，一个升级Europe PMC接口的知识层，面向AI agent，提升多项任务表现：文献综合、claim验证、开放域问答，以及下游生物学任务如protocol问题与序列操作。EMBL AI Librarian is introduced, a knowledge layer that upgrades the Europe PMC interface for AI agents that improves performance across a range of tasks: literature synthesis, claim verification, open-domain question answering, and downstream biology tasks such as protocol questions and sequence manipulation.

训练后对齐往往较浅，会在微调中被侵蚀。而中训练干预能否在干净隔离于训练后的情况下产生持久对齐，此前未经检验。我们通过宪法式中训练来测试：在 120B 规模上，插入基于原则与价值观的内容，与仅做回放的对照组进行对比。我们基于 Anthropic 的 Constitution 构建了 394M token 的宪法语料，并采用 2×2 析因设计（课程顺序 × 审慎推理），形成四种宪法式中训练条件与一组对照，随后在自生成与既有...Post-training alignment is often shallow, eroding under fine-tuning. Whether midtraining interventions, cleanly isolated from post-training, can produce durable alignment remains untested. We test this via constitutional midtraining: inserting principled, values-based content into midtraining against a replay-only control at 120B scale. Our 394M-token constitutional corpus, built from Anthropic's Constitution, uses a 2x2 factorial design (curriculum ordering x deliberative reasoning) to produce four constitutionally midtrained conditions plus a control, evaluated on self-generated and establis

UEmbed (Unified Embedding)是一种decoder-only多模态嵌入模型，在单次因果前向中同时产出稀疏词项与稠密表示，提供新范式：在单一模型中统一稠密与稀疏嵌入，并将稀疏检索扩展以统一文本与多模态输入。UEmbed (Unified Embedding), a decoder-only multimodal embedding model that produces both sparse lexical and dense representations in one causal forward pass, offers a new paradigm: it unifies dense and sparse embeddings in one model, while further extending sparse retrieval to unify text and multimodal inputs.

将长视野执行重新表述为任务状态管理问题，提出LongHorizon-Harness，在执行外部显式维护任务状态，并仅用从环境中独立验证的事实更新它。This work reformulate long-horizon execution as a task-state management problem and proposes LongHorizon-Harness, which maintains the task state explicitly outside execution and updates it only with facts independently verified from the environment.

本文将技能生成建模为序列编辑过程，把技能构建分解为可单独评估的编辑，并提出一种新颖的回滚奖励，通过在锚定查询上对比原始技能与编辑后技能的下游执行效果来评估每次编辑。This work forms skill generation as a sequential editing process that decomposes skill construction into individually evaluable edits, and introduces a novel rollback reward that evaluates each edit by comparing downstream execution under the original and edited skills on an anchored query.

本文提出 Motion Beyond Morphology（超越形态的运动）这一视角，旨在跨固定结构对应迁移运动，通过两阶段框架保留在不同目标形态间仍具有意义的动力学。This work introduces Motion Beyond Morphology, a perspective that seeks to transfer motion beyond fixed structural correspondence, by preserving dynamics that remain meaningful across different target morphologies by proposing a two-stage framework.

本文提出 DAPD，一种具有两级锚定的统一框架，可显著缓解特权错觉，在 Qwen3-4B 上以平均 +2.00 分优于 OPSD。DAPD is proposed, a unified framework with two levels of anchoring that significantly alleviates privilege illusion, outperforming OPSD on Qwen3-4B by +2.00 points on average across tasks.

本文提出 3DZip，一种三阶段 token 压缩框架：首先采用粗粒度体素化去除点级冗余，再通过 Determinantal Point Process 基于特征空间多样性选取锚点 token，最后在空间约束下融合剩余 token 以保持几何一致性。3DZip is proposed, a three-stage token compression framework that first applies coarse voxelization to remove point-level redundancy, then selects anchor tokens based on feature-space diversity via a Determinantal Point Process, and finally merges remaining tokens under spatial constraints to preserve geometric coherence.

本文提出 CADENA（西班牙语意为"链"），一种将 3D 网格重建为参数化 CAD 程序的模型，按顺序逐个生成操作序列，并在每一步将目标与当前预测几何进行对比。This work introduces CADENA (Spanish for"chain"), a model that reconstructs a 3D mesh as a parametric CAD program, growing its sequence of operations one at a time and comparing the target with the currently predicted geometry at every step.

本文提出 LeapTalk，一种通过单次前向实现稳定且实时说话头生成、可扩展至任意长视频的新颖框架，并引入音频驱动的无分类器引导机制，在极端步数缩减下保持细粒度唇形同步。LeapTalk is proposed, a novel framework that achieves stable and real-time talking-head generation with a single forward step, scaling to arbitrarily long videos, and an audio-driven classifier-free guidance mechanism to maintain fine-grained lip synchronization under extreme step reduction.

GradCuit（梯度穿越电路）在所选 Transformer 层、提示隐藏表示与生成续写之间插入可优化的潜变量，开启了鲁棒且可解释的测试时缩放新维度，使 LLM 调整其推理方式，而不仅仅是重新生成、采样或重排输出。GradCuit (gradient through circuit), which inserts optimizable latent states at a selected Transformer layer between the hidden representations of the prompt and the generated continuation, opens a new axis of robust and interpretable test-time scaling, where LLMs adapt how they reason rather than merely regenerate, sample, or rerank outputs.

本文提出 DeepVoyager-VL，一种面向视觉在环搜索的长程多模态深度搜索框架，通过构建多模态事件图驱动数据合成，从而产出具有中间视觉依赖与长推理链的问题。DeepVoyager-VL is proposed, a long-horizon multimodal deep-search framework for vision-in-the-loop search that constructs a multimodal event graph to drive data synthesis, yielding problems with intermediate visual dependencies and long reasoning chains.

本文提出 DreamTraj，从单张 RGB 图像和任务指令预测物体 6-DoF 轨迹，推理时无需视频、深度或 CAD 模型，是首个直接从中间视频扩散表征（而非生成像素）解码物体 6-DoF 轨迹的方法This work proposes DreamTraj, which predicts a 6-DoF object trajectory from a single RGB image and a task instruction, requiring no video, depth, or CAD model at inference, and is the first approach to directly decode object 6-DoF trajectories from intermediate video diffusion representations rather than generated pixels.

在四种拆分 backbone 上，ReBA 降低所有报告 benchmark 输入的负载，同时保持与 Std-Aux 相当的平均任务准确率，并在分辨率与分块变化下降低测试范围内的平均负载与最差物理负载Across four split backbones, ReBA lowers load on every reported benchmark input while keeping mean task accuracy comparable to Std-Aux, and lowers average load over the tested range and worst physical load under resolution and tiling shifts.

RecHarness 将优化过程分为两步：bandit 路由器根据历史验证反馈选择下一步修改方向，LLM 在选定方向内生成具体优化假设与可执行代码编辑RecHarness separates the optimization process into two steps: a bandit router selects the next modification direction according to historical validation feedback, while the LLM generates a concrete optimization hypothesis and executable code edit within the selected direction.

本文在中间层附加轻量预测头，保持 backbone 冻结，并利用中间预测与最终预测的差异作为采样时的自引导方向，训练一个能够自引导的冻结预训练像素扩散模型This work attaches a lightweight prediction head to an intermediate layer, keep the backbone frozen, and use the discrepancy between the intermediate and final predictions as a self-guidance direction during sampling to train a frozen, pretrained pixel diffusion model that can guide itself.

本文提出 GPTQ-2D，以三次时间复杂度生成相同的取整矩阵，并研究该任务的双侧版本——固定非奇异基矩阵同时作用于残差的左右两侧This work presents GPTQ-2D, which produces the identical rounded matrix in cubic time, and studies the two-sided version of this task, in which fixed nonsingular basis matrices act on both the left and the right of the residual.

本文提出 Wnuan 三阶段流程：从文档构建任务导向监督，结合通用数据回放进行监督微调，对残余误差应用强化学习，阐述分阶段企业适配的收益与通用能力代价Wnuan, a three-stage pipeline that constructs task-oriented supervision from documents, performs supervised fine-tuning with general-data replay, and applies reinforcement learning to residual errors is presented, describing both the gains and the general-capability cost of staged enterprise adaptation.

结果表明结构化 agent 记忆无需生成过去的中间表示，Zero-Mem 在消除记忆操作中 LLM 调用与 LLM-token 消耗的同时取得具有竞争力的性能The results show that structured agent memory need not generate an intermediate representation of the past, and Zero-Mem achieves competitive performance while eliminating LLM calls and LLM-token consumption from memory operations.

在后训练中教授删除操作可减少删除回避行为并提升更广泛的代码编辑性能，表明该行为是训练不足而非不可达成Teaching deletion during post-training reduces deletion avoidance and improves broader code-editing performance, suggesting the behavior is undertrained rather than beyond reach.

将 LLM 适配到专用领域常会带来对齐税：针对领域特定任务进行微调会导致灾难性遗忘，并显著降低在通用任务上的表现。我们提出 MemSFT，通过将领域专业化与主干参数更新解耦，以即插即用的参数化记忆来缓解对齐税。该记忆被训练为模仿在领域数据上运作的非参数化检索器，从而记住原本需通过检索获取的知识与模式。一旦在Adapting Large Language Models (LLMs) to specialized domains often incurs an alignment tax, as fine-tuning on domain-specific tasks can cause catastrophic forgetting and substantially degrade performance on general tasks. We propose MemSFT, which mitigates the alignment tax by decoupling domain specialization from backbone parameter updates through a plug-and-play parametric memory. The memory is trained to imitate the behavior of a non-parametric retriever operating over domain data, thereby memorizing knowledge and patterns that would otherwise be accessed through retrieval. Once trained on

在作者研究的粗粒度属性上，MLLM 编码了视觉证据但无法可靠控制对其的依赖For the coarse attributes the authors study, MLLMs encode the visual evidence but cannot reliably control their reliance on it, indicating that for the coarse attributes they study, MLLMs cannot reliably control their reliance on it.

结果是面向稀疏 event-KV 服务的记忆契约：写入什么、落在何处、源消失后什么得以保留The result is a memory contract for sparse event-KV serving: what to write, where it lands, and what survives once the source is gone.

提出 Video-DR，采用解耦的感知-探索流水线与分阶段工具解锁，强制在 web 检索前进行充分的跨帧视觉定位，实现突破模仿学习上限的自主探索。Video-DR is introduced, featuring a decoupled perception-exploration pipeline with stage-wise tool unlocking that compels exhaustive cross-frame visual grounding prior to web retrieval, enabling autonomous exploration that breaks the imitation-learning ceiling.

提出 Knowledge-Geometry Decoupling (KGD) 并引入 Behavioral Multi-Token Prediction (BMTP)，仅将协作或语义相关的未来项作为监督，从而得到更干净、更可迁移的行为知识。Knowledge-Geometry Decoupling (KGD) is proposed and Behavioral Multi-Token Prediction (BMTP) is introduced to retain only collaboratively or semantically related future items as supervision, yielding cleaner and more transferable behavioral knowledge.

将 Agent-Centric Interactive World Proxies 概念化，将基础范式从物理状态转移转向 agent 可用的信息转移，如执行结果、检索到的经验或技能、以及验证信号，扩展了世界建模的范围，为持续改进的 agent 提供多样化反馈。This work conceptualize Agent-Centric Interactive World Proxies, shifting the fundamental paradigm from physical state transitions to agent-usable information transitions, such as execution outcomes, retrieved experiences or skills, and verification signals, broadening the scope of world modeling to provide versatile feedback for continually improving agents.

高粘度污渍以其高粘度与复杂流变特性，仍是机器人表面清洁的主要挑战。传统擦拭往往扩散污渍，而擦洗摩擦力更强却存在损伤表面的风险。本文提出 Push-Wiper，一种将高粘度污渍清洁重构为聚集问题的框架。Push-Wiper 使用海绵通过分段推送轨迹渐进式聚集污渍，随后通过后处理阶段剥离已聚集物质并实现海绵自清洁。我们采用逐步Viscous stains, characterized by high viscosity and complex rheological properties, remain a major challenge for robotic surface cleaning. Conventional wiping often spreads the stain, while scrubbing provides stronger friction but risks damaging the surface. In this paper, we propose Push-Wiper, a framework that reformulates viscous stain cleaning as an aggregation problem. Push-Wiper employs a sponge to progressively gather stains through segmented pushing trajectories, followed by a post-processing phase that detaches the aggregated material and enables sponge self-cleaning. We adopt a stepw

提出 Semantic-Temporal WAM (ST-WAM)，使用 DINOv3 作为未来预测与历史检索的共享语义表示，同时保留细粒度 VAE 动力学，以提升动作鲁棒性；证明语义-时间建模能有效补充像素生成动力学，实现稳健的操作。Semantic-Temporal WAM (ST-WAM) is proposed to improve action robustness by using DINOv3 as a shared semantic representation for future prediction and history retrieval while retaining fine-grained VAE dynamics, demonstrating that semantic-temporal modeling effectively complements pixel-generative dynamics for robust manipulation.

提出 ContinualSkillBench，一个面向 in-context 持续 skill 学习的动态评估框架，表明当前 in-context skill 进化机制能够支持持续适应，但仍难以稳定地将经验整合为鲁棒且可迁移的 skill。ContinualSkillBench is introduced, a dynamic evaluation framework for in-context continual skill learning that shows that current in-context skill evolution mechanisms can support continual adaptation, but still struggle to consistently consolidate experience into robust and transferable skills.