论文

本文倡导源对比评估，并构建了 Cultivar——FLORES 的本地化子集，可用于特定 locale 的翻译评估；研究发现：MT 专用模型鲁棒性较差，少数模型可能对 FLORES 存在过拟合，且模型普遍更擅长翻译美国 locale 的内容，而非其他 locale，无论语种如何。This work advocates for source-contrastive evaluation and instantiates Cultivar, a localised subset of FLORES, which enables locale-specific translation evaluation and finds that MT-specialised models are less robust, a few models potentially overfit FLORES, and models tend to translate US content better than that of other locales, regardless of language.

本文提出 Business Arena——一个受控环境，AI agent 在其中经营跨境店铺，在长周期内向供应商采购并向买家销售，迈出了构建面向端到端商业 agent 的真实可信测试床的第一步。Business Arena, a controlled environment where an AI agent runs a cross-border shop, buying from suppliers and selling to buyers over a long horizon, is introduced, taking a first step toward a realistic and trustworthy testbed for evaluating end-to-end business agents.

面向策略性优化下的可测性设计指南：保留探针仅在不可枚举轴上保持有效性；门控必须衡量保留集性能，而非仅正确性；迁移率只有在附带每类失败机制评级时才可解读。Design guidance for measurement under strategic optimization is distill design guidance for measurement under strategic optimization: held-out probes retain validity only on non-enumerable axes; gates must measure held-out performance, not just correctness; and a transfer rate is interpretable only with per-failure mechanism grades.

本文实现了对时间序列数据集相似度方法的系统化、可复现比较，提供灵活的扩展性以添加自定义数据集、相似度方法及下游时间序列任务，并通过集成的时间序列数据集归约器对数据集级和序列级相似度方法进行一致评估。This work enables systematic and reproducible comparisons of time-series dataset similarity methods, flexible extensibility for users to add customized datasets, similarity methods, and downstream time-series tasks, and consistent evaluation of both dataset-level and series-level similarity methods through integrated time-series dataset reducers.

本文提出解码层禁忌 (Decoding-Level Taboo)，一种零提示的诊断式压力测试，在运行时直接干预 logit 空间，于词边界处动态遮蔽主要候选 token，强制模型进行迂回表达。Decoding-Level Taboo is introduced, a zero-prompt diagnostic stress test that intervenes directly in logit space at runtime, forcing models out of their nominal paths by dynamically masking primary candidate tokens at word boundaries, forcing machine circumlocution.

360CityArena 为真实感城市区域导航与空间推理提供了必要且具有挑战性的测试平台；基于 SOTA LMM 智能体的评估显示，即使是最强的模型 Gemini 2.5 Flash，其表现仍远低于人类水平。360CityArena provides a necessary and challenging testbed for photorealistic urban-district navigation and spatial reasoning, and evaluation using state-of-the-art LMM-based agents shows that even the strongest model, Gemini 2.5 Flash, performs far below human level.

本文将该挑战形式化为叙事承诺保持 (Narrative Commitment Preservation, NCP)，并提出 NCP-Bench：一个基于电影剧情梗概构建的包含 100 个叙事环境的基准，每个环境均提供可在玩家智能体与叙述者智能体交互过程中自动检查的结构化叙事规范。This work forms this challenge as Narrative Commitment Preservation (NCP), and introduces NCP-Bench, a benchmark of 100 narrative environments derived from movie synopses that each environment includes a structured narrative specification that can automatically check throughout the interaction between the player agent and the narrator agent.

本文提出进化式马尔可夫超图攻击（EMHA），这是一种黑盒策略，通过协调授权状态转移执行反馈驱动的环境演化，无需参数更新，并将 OpenART 确立为在复杂演化环境中研究 Agent 安全性的可扩展基础。This work proposes the Evolutionary Markov Hypergraph Attack (EMHA), a black-box policy that performs feedback-driven environment evolution by coordinating authorized state transitions without requiring parameter updates, and establishes OpenART as a scalable foundation for studying agent safety in complex, evolving environments.

H2R-Bench 提供了一个系统性诊断框架，用于评估视频世界模型能否跨越 human-to-robot 具身差距，并将人类操作观测转化为以机器人为中心的训练资源。H2R-Bench provides a systematic diagnostic framework for evaluating whether video world models can bridge the human-to-robot embodiment gap and convert human manipulation observations into robot-centric training resources.

基于 CPI-Bench 对主流图像编辑模型的评测结果显示，CPI-Bench 增强了模型间的性能区分度；排名分析表明 CPI-Bench 与 Arena Image Edit Leaderboard 的对齐度最高，与公开人类偏好排名具有更强的一致性。Evaluation results of mainstream image editing models based on CPI-Bench demonstrate that CPI-Bench enhances performance differentiation among models, and ranking analysis reveals that CPI-Bench achieves the highest alignment with the Arena Image Edit Leaderboard, indicating stronger consistency with public human preference rankings.

在对 1,000 只美股进行小时级收益预测时，我们观察到一种意外现象：预测结果近乎平坦，且截面相关性衡量的股票排序能力很差，我们将其称为"预测坍缩"。令人惊讶的是，在同一设置下对交易量进行预测时，该现象基本消失。我们在多种时序基础模型（TSFMs）、12 个深度学习预测模型以及 97 个公开基准配置中系统考察了这一现象，发现其与目标可预测性密切相关，并识别出背后的两类成因：低 p……When forecasting hourly returns for 1,000 US equities, we observe an unexpected phenomenon: predictions become nearly flat and show poor stock ranking, as measured by cross-sectional correlation. We call this forecast collapse. Surprisingly, the phenomenon largely disappears when forecasting trading volume under the same setting. We investigate forecast collapse across time-series foundation models (TSFMs), twelve deep-learning forecasting models, and 97 public benchmark configurations, and find that it is closely tied to target predictability. We identify two distinct reasons behind it: low p

一个机器人过程评估工具包，将 rollout 视频转化为稠密进度曲线并衍生多项细粒度指标；引入 RoboPulse++ 用于评估过程奖励模型（PRM）的可靠性，为评测者提供更准确的测试平台。A toolkit for robot process assessment that turns rollout videos into dense progress curves and derives multiple fine metrics, and introduces RoboPulse++ to evaluate the reliability of process reward models (PRM), providing evaluators with a more accurate testing platform.

提出 MobileMem，一个面向设备端长期记忆研究的 benchmark 和框架，基于长达一年的移动端经验集合，使 agent 能够记忆过去、理解当下并适应未来。This work introduces MobileMem, a benchmark and framework for studying on-device long-term memory, grounded in a year-scale collection of mobile experiences, and enables agents to remember the past, understand the present, and adapt to the future.

提出 UNMASK，一个全自动 pipeline，可在无需额外人工标注的情况下发现、因果验证并缓解文本分类器中的伪相关，并证明其发现与验证阶段可泛化至奖励模型的偏好数据。U N M ASK is presented, a fully automated pipeline that discovers, causally verifies, and mitigates spurious correlations in text classifiers without additional human annotation, and demonstrates that the discovery and validation stages generalize to reward model preference data.

发现面向推理的训练并未优先放大具有最高 Lift 的行为，这促使研究者采用过程级目标，以奖励经过校准且有依据的推理，而不仅仅是表面形式。It is found that reasoning-oriented training does not preferentially amplify the highest-Lift behaviors, motivating process-level objectives that reward calibrated and grounded reasoning rather than surface form alone.

提出 Apodex Discovery，一个通过 heavy-duty solver 构建和评估发现型 AI 的框架；该 solver 包含一个 foundation model、harness、工具和控制策略，用于执行长期的、有状态的、可验证的探索。This work introduces Apodex Discovery, a framework for building and evaluating discoverative AI through the heavy-duty solver, a system comprising a foundation model, harness, tools, and control policies that pursues extended, stateful, verifiable investigations.

探讨共享工作空间上 LLM Agent 委员会的审议过程是否能被捷径和线索（benchmark 所奖励但临床医生会忽略的）所博弈，以及委员会的社会可信度所构成的游戏。It is asked whether committees of language-model agents deliberating on a shared workspace can be gamed by shortcuts, cues a benchmark rewards but a clinician would ignore, and what games a committee is social plausibility.

提出 GRIP（Grounded Reasoning via Information-Restricted Premises），引入容量不对称：decoder 对 query 保持全维度访问，而检索到的证据则通过一个严苛的随机瓶颈，迫使证据通道仅编码 query 中无法获得的残余信息。GRIP (Grounded Reasoning via Information-Restricted Premises), which imposes capacity asymmetry: the decoder keeps full-dimensional access to the query, while retrieved evidence passes through a severe stochastic bottleneck, which forces the evidence channel to encode only the residual information unavailable from the query.

提出 HarnessEval-W，一个 agentified 的评估 pipeline，将 LLM 生态中的 harness 范式引入 world model 基准测试，并在 330 个评估用例上对 18 个代表性 world model 进行了评估。This work introduces HarnessEval-W, an agentified evaluation pipeline that brings the harness paradigm from the LLM ecosystem to world model benchmarking, and applies HarnessEval-W to 18 representative world models over 330 evaluation cases.

语言模型以一种可被报告的形式持有潜在量，并且当任务需要灵活复用该量时，更多该量的信息会以这种形式存在。Language models hold latent quantities in a form they can report on, and more of a quantity is present in that form when the task requires reusing it flexibly when the task requires reusing it flexibly.

认识到多样化的多参考任务共享一组共同的原子操作，本文形式化了四个算子：Anchor、Disentangle、Apply 和 Compose，并构建了 TRACE-Bench，包含约 1,600 个跨 slot 数量 1–8 的评估用例。Recognizing that diverse multi-reference tasks share a common set of atomic operations, this work formalizes four operators: Anchor, Disentangle, Disentangle, Apply, and Compose, and constructs TRACE-Bench, comprising approximately 1,600 evaluation cases across slot counts 1--8.