评分理由: 论文QuarkMedSearch专注于医疗领域的深度搜索智能体研究，基于基础模型Tongyi DeepResearch构建。核心相关关键词包括：1) “Large Language Models” (权重1.0) - 论文明确基于"agentic foundation models"，属于大模型应用，给10分；2) “Post-training” (权重1.0) - 论文采用两阶段SFT和RL训练策略，是核心方法，给10分；3) “RLHF” (权重1.0) - 论文提到RL训练，与RLHF相关但非核心，给5分；4) “Chain of Thought” (权重1.0) - 深度搜索涉及多步推理，给5分；5) “System 2 Thinking” (权重1.0) - 深度搜索需要深入推理，给5分；6) “Self-Correction” (权重1.0) - 论文强调反思能力，给10分；7) “LLM Agents” (权重1.0) - 论文是智能体研究，给10分；8) “Tool Use” (权重1.0) - 论文涉及工具调用，给10分；9) “AI for Science” (权重1.0) - 医疗AI应用，给10分。其他关键词如MoE、量化、RAG等未在摘要中体现，给0分。

随着具身化基础模型的持续演进，如何进一步提升其在垂直领域中的性能已成为重要挑战。为此，我们在强大的具身化基础模型通义DeepResearch基础上，聚焦中文医疗深度搜索场景，提出QuarkMedSearch，系统性地探索了涵盖医疗多跳数据构建、训练策略与评估基准的全流程方法，以进一步推进并评估其在垂直领域的性能上限。具体而言，在数据合成方面，针对医疗领域深度搜索训练数据稀缺的问题，我们结合大规模医疗知识图谱与实时在线探索，构建了长周期的医疗深度搜索训练数据；在训练后优化阶段，我们采用两阶段的监督微调（SFT）与强化学习（RL）训练策略，逐步增强模型进行深度搜索所需的规划、工具调用与反思能力，同时保持搜索效率；在评估方面，我们与医学专家合作，通过严格的人工验证构建了QuarkMedSearch Benchmark。实验结果表明，QuarkMedSearch在QuarkMedSearch Benchmark上取得了同规模开源模型中的最优性能，同时在通用基准测试中也保持了强劲的竞争力。

As agentic foundation models continue to evolve, how to further improve their performance in vertical domains has become an important challenge. To this end, building upon Tongyi DeepResearch, a powerful agentic foundation model, we focus on the Chinese medical deep search scenario and propose QuarkMedSearch, systematically exploring a full-pipeline approach spanning medical multi-hop data construction, training strategies, and evaluation benchmarks to further push and assess its performance upper bound in vertical domains. Specifically, for data synthesis, to address the scarcity of deep search training data in the medical domain, we combine a large-scale medical knowledge graph with real-time online exploration to construct long-horizon medical deep search training data; for post-training, we adopt a two-stage SFT and RL training strategy that progressively enhances the model’s planning, tool invocation, and reflection capabilities required for deep search, while maintaining search efficiency; for evaluation, we collaborate with medical experts to construct the QuarkMedSearch Benchmark through rigorous manual verification. Experimental results demonstrate that QuarkMedSearch achieves state-of-the-art performance among open-source models of comparable scale on the QuarkMedSearch Benchmark, while also maintaining strong competitiveness on general benchmarks.

评分理由: 论文核心研究LLM-based多模态情感识别，提出AffectAgent框架，包含三个协作智能体（query planner, evidence filter, emotion generator），使用Multi-Agent Proximal Policy Optimization (MAPPO)优化。高度相关的关键词包括：LLMs（核心基础）、MoE（提出MB-MoE方法）、RAG（检索增强生成框架）、LLM Agents和Multi-agent Systems（多智能体协作框架）。中等相关的关键词：Chain of Thought和System 2 Thinking（涉及分析推理过程）、Hallucination Mitigation（解决幻觉问题）。其他关键词如SLMs、Scaling Laws、Pre-training等未在论文中涉及。

基于大语言模型的多模态情感识别依赖静态参数化记忆，在解读细微情感状态时常产生幻觉。本文针对单轮检索增强生成方法易受模态模糊性影响、难以捕捉跨模态复杂情感依赖的问题，提出AffectAgent——一种面向情感的多智能体检索增强生成框架，该框架利用智能体间的协同决策实现细粒度情感理解。具体而言，AffectAgent包含三个联合优化的专用智能体：查询规划器、证据过滤器和情感生成器，它们通过协作式分析推理实现跨模态样本检索、证据评估与预测生成。这些智能体采用多智能体近端策略优化算法进行端到端优化，并通过共享情感奖励机制确保情感理解的一致性。此外，我们提出模态平衡专家混合模块与检索增强自适应融合模块：前者动态调节不同模态的贡献以缓解跨模态异质性导致的表征失配，后者通过引入检索到的视听嵌入向量增强缺失模态条件下的语义补全能力。在MER-UniBench基准上的大量实验表明，AffectAgent在复杂场景中均取得优越性能。代码发布于：https://github.com/Wz1h1NG/AffectAgent。

LLM-based multimodal emotion recognition relies on static parametric memory and often hallucinates when interpreting nuanced affective states. In this paper, given that single-round retrieval-augmented generation is highly susceptible to modal ambiguity and therefore struggles to capture complex affective dependencies across modalities, we introduce AffectAgent, an affect-oriented multi-agent retrieval-augmented generation framework that leverages collaborative decision-making among agents for fine-grained affective understanding. Specifically, AffectAgent comprises three jointly optimized specialized agents, namely a query planner, an evidence filter, and an emotion generator, which collaboratively perform analytical reasoning to retrieve cross-modal samples, assess evidence, and generate predictions. These agents are optimized end-to-end using Multi-Agent Proximal Policy Optimization (MAPPO) with a shared affective reward to ensure consistent emotion understanding. Furthermore, we introduce Modality-Balancing Mixture of Experts (MB-MoE) and Retrieval-Augmented Adaptive Fusion (RAAF), where MB-MoE dynamically regulates the contributions of different modalities to mitigate representation mismatch caused by cross-modal heterogeneity, while RAAF enhances semantic completion under missing-modality conditions by incorporating retrieved audiovisual embeddings. Extensive experiments on MER-UniBench demonstrate that AffectAgent achieves superior performance across complex scenarios. Our code will be released at: https://github.com/Wz1h1NG/AffectAgent.

评分理由: 论文提出LoSA方法，针对块状扩散语言模型（DLMs）在长上下文场景中的内存瓶颈问题，通过利用令牌隐藏状态的稳定性，设计了一种局部感知的稀疏注意力机制。该方法核心涉及稀疏模型（MoE/Sparse Models）和注意力优化技术（KV Cache Compression/Linear Attention/FlashAttention），以解决长上下文（Long Context LLMs）处理中的效率问题，并实现推理加速（Inference Acceleration）。论文虽未直接使用"LLMs"术语，但DLMs属于大语言模型的一种变体，因此给予中等相关度。其他关键词如小模型、训练方法、对齐、代理等与论文内容无关。

分块扩散语言模型（DLMs）能够以任意顺序生成多个词元，为自回归解码流程提供了一种有前景的替代方案。然而，在长上下文场景中，它们仍然受限于内存约束的注意力机制。由于存在KV膨胀问题，朴素的稀疏注意力在DLMs上效果不佳：不同的查询会选择不同的前缀位置，导致被访问的KV页面集合过大。为解决此问题，我们观察到在连续的降噪步骤之间，只有一小部分活跃词元的隐藏状态发生显著变化，而大多数稳定词元的状态几乎保持不变。基于这一洞察，我们提出了局部感知稀疏注意力（Locality-aware Sparse Attention, LOSA）。该方法对稳定词元复用已缓存的前缀注意力结果，仅对活跃词元应用稀疏注意力。这大幅减少了必须加载的KV索引数量，从而实现了更高的加速比和更高的准确性。在多种分块DLMs和基准测试中，LOSA在保持接近稠密模型准确性的同时，显著提升了效率：在激进的稀疏度水平下，平均准确率最高提升9个百分点，同时注意力密度降低至1.54倍以下。在RTX A6000 GPU上，注意力计算速度最高提升4.14倍，证明了所提方法的有效性。

Block-wise diffusion language models (DLMs) generate multiple tokens in any order, offering a promising alternative to the autoregressive decoding pipeline. However, they still remain bottlenecked by memory-bound attention in long-context scenarios. Naive sparse attention fails on DLMs due to a KV Inflation problem, where different queries select different prefix positions, making the union of accessed KV pages large. To address this, we observe that between consecutive denoising steps, only a small fraction of active tokens exhibit significant hidden-state changes, while the majority of stable tokens remain nearly constant. Based on this insight, we propose LOSA (Locality-aware Sparse Attention), which reuses cached prefix-attention results for stable tokens and applies sparse attention only to active tokens. This substantially shrinks the number of KV indices that must be loaded, yielding both higher speedup and higher accuracy. Across multiple block-wise DLMs and benchmarks, LOSA preserves near-dense accuracy while significantly improving efficiency, achieving up to +9 points in average accuracy at aggressive sparsity levels while maintaining 1.54x lower attention density. It also achieves up to 4.14x attention speedup on RTX A6000 GPUs, demonstrating the effectiveness of the proposed method.

评分理由: 论文核心研究LLMs处理长上下文时的效率瓶颈（KV缓存和计算复杂度），提出Latent-Condensed Attention（LCA）方法。高度相关（10分）的关键词：“Large Language Models”（论文明确针对LLMs）、“Context Window Extension”（解决长上下文建模）、“KV Cache Compression”（核心贡献是减少KV缓存）。中等相关（5分）的关键词：“Mixture of Experts”（提及稀疏方法作为背景）、“Quantization”（涉及模型效率优化）、“Speculative Decoding”（属于推理加速范畴）。其他关键词与论文内容无直接关联（0分）。

大语言模型在处理长上下文时面临显著挑战，这主要源于键值缓存（KV cache）的线性增长与自注意力机制（self-attention）的二次计算复杂度。现有方法通常分别应对这些瓶颈：多头潜在注意力（Multi-head Latent Attention, MLA）通过将词元投影至低维潜在空间以减少KV缓存，而稀疏注意力则致力于降低计算开销。然而，稀疏方法无法直接在MLA压缩后的潜在结构上运行，错失了联合优化的机会。本文提出潜在凝聚注意力（Latent-Condensed Attention, LCA），该方法直接在MLA的潜在空间内对上下文进行凝聚，其中表示被解耦为语义潜在向量与位置键。LCA通过查询感知池化分别聚合语义向量，并通过锚点选择保留位置键。这一设计在不增加参数的前提下，联合降低了计算成本与KV缓存。除MLA外，LCA的设计与架构无关，可轻松扩展至其他注意力机制，如分组查询注意力（GQA）。理论上，我们证明了其误差界与长度无关。实验表明，在128K上下文长度下，LCA实现了高达2.5倍的预填充加速与90%的KV缓存减少，同时保持了具有竞争力的性能。

Large language models (LLMs) face significant challenges in processing long contexts due to the linear growth of the key-value (KV) cache and quadratic complexity of self-attention. Existing approaches address these bottlenecks separately: Multi-head Latent Attention (MLA) reduces the KV cache by projecting tokens into a low-dimensional latent space, while sparse attention reduces computation. However, sparse methods cannot operate natively on MLA’s compressed latent structure, missing opportunities for joint optimization. In this paper, we propose Latent-Condensed Attention (LCA), which directly condenses context within MLA’s latent space, where the representation is disentangled into semantic latent vectors and positional keys. LCA separately aggregates semantic vectors via query-aware pooling and preserves positional keys via anchor selection. This approach jointly reduces both computational cost and KV cache without adding parameters. Beyond MLA, LCA’s design is architecture-agnostic and readily extends to other attention mechanisms such as GQA. Theoretically, we prove a length-independent error bound. Experiments show LCA achieves up to 2.5$\times$ prefilling speedup and 90% KV cache reduction at 128K context while maintaining competitive performance.

评分理由: 论文提出HypoExplore框架，使用大语言模型（LLM）作为核心组件来生成和评估神经网络架构假设，属于LLM在科学发现（AI for Science）领域的创新应用。框架本质上是基于LLM的自主智能体（LLM Agents）系统，通过多智能体（Multi-agent Systems）协作分析实验结果，并涉及假设驱动的推理过程（Chain of Thought/System 2 Thinking）和自我改进机制（Self-Correction）。论文主要聚焦于视觉架构发现，未涉及其他关键词如MoE、训练技术、推理优化等具体技术细节。

我们提出HypoExplore，这是一个智能体框架，它将视觉识别中的神经架构发现构建为一种假设驱动的科学探究过程。在给定人类指定的高层次研究方向后，HypoExplore通过演化分支进行神经架构的构思、实现、评估与改进。新假设的生成利用大型语言模型完成，通过选择父假设作为构建基础，并遵循一种双重策略进行引导：该策略既注重利用已验证的设计原则，也致力于解决尚不确定的问题。我们提出的框架维护着一个记录所有提出架构谱系的“轨迹树”，以及一个通过实验证据主动追踪置信度的“假设记忆库”。每次实验后，多个反馈智能体会从不同角度分析结果，并将其发现整合为假设置信度的更新。我们在CIFAR-10数据集上测试了该框架以发现轻量级视觉架构，其中最佳架构从初始基线精度18.91%演化至94.11%的准确率，并成功推广至CIFAR-100和Tiny-ImageNet。我们通过在MedMNIST数据集上进行独立的架构发现实验，进一步证明了该框架在专业领域的适用性，所获架构达到了最先进的性能。研究表明，随着证据的积累，假设置信度分数变得越来越具有预测性，并且学习到的设计原则能够在独立的演化谱系间迁移，这表明HypoExplore不仅能发现更强的架构，还有助于真正理解设计空间。

We introduce HypoExplore, an agentic framework that formulates neural architecture discovery for visual recognition as a hypothesis-driven scientific inquiry. Given a human-specified high-level research direction, HypoExplore ideates, implements, evaluates, and improves neural architectures through evolutionary branching. New hypotheses are created using a large language model by selecting a parent hypothesis to build upon, guided by a dual strategy that balances exploiting validated principles with resolving uncertain ones. Our proposed framework maintains a Trajectory Tree that records the lineage of all proposed architectures, and a Hypothesis Memory Bank that actively tracks confidence scores acquired through experimental evidence. After each experiment, multiple feedback agents analyze the results from different perspectives and consolidate their findings into hypothesis confidence updates. Our framework is tested on discovering lightweight vision architectures on CIFAR-10, with the best achieving 94.11% accuracy evolved from a root node baseline that starts at 18.91%, and generalizes to CIFAR-100 and Tiny-ImageNet. We further demonstrate applicability to a specialized domain by conducting independent architecture discovery runs on MedMNIST, which yield a state-of-the-art performance. We show that hypothesis confidence scores grow increasingly predictive as evidence accumulates, and that the learned principles transfer across independent evolutionary lineages, suggesting that HypoExplore not only discovers stronger architectures, but can help build a genuine understanding of the design space.

评分理由: 论文核心研究RAG方法创新，与"Retrieval-Augmented Generation"高度相关（15分），直接应用于大语言模型（10分）。论文关注顺序敏感推理，与"Chain of Thought"和"System 2 Thinking"有一定关联（各8分），涉及多步推理和深度推理过程。其他关键词如MoE、量化、对齐等未在论文中涉及，得0分。

检索增强生成（RAG）通过将输出建立在检索到的知识基础上，增强了大型语言模型的能力。然而，现有的RAG方法，包括基于图和超图的方法，都将检索到的证据视为无序集合，隐含地假设了排列不变性。这一假设与许多现实世界的推理任务不符，因为任务结果不仅取决于发生了哪些交互，还取决于这些交互展开的顺序。我们提出了顺序感知知识超图RAG（Order-Aware Knowledge Hypergraph RAG, OKH-RAG），它将顺序视为一种首要的结构属性。OKH-RAG将知识表示为带有优先序结构的超图中的高阶交互，并将检索重新定义为对超边的序列推断。它并非选择独立的事实，而是恢复反映底层推理过程的连贯交互轨迹。一个学习到的转移模型直接从数据中推断优先序，而无需显式的时间监督。我们在顺序敏感的问答和解释任务（包括热带气旋和港口运营场景）上评估了OKH-RAG。OKH-RAG始终优于排列不变的基线方法，消融实验表明，这些性能提升正是源于对交互顺序的建模。这些结果突显了基于集合的检索的一个关键局限：有效的推理不仅需要检索相关证据，还需要将其组织成结构化的序列。

Retrieval-augmented generation (RAG) enhances large language models by grounding outputs in retrieved knowledge. However, existing RAG methods including graph- and hypergraph-based approaches treat retrieved evidence as an unordered set, implicitly assuming permutation invariance. This assumption is misaligned with many real-world reasoning tasks, where outcomes depend not only on which interactions occur, but also on the order in which they unfold. We propose Order-Aware Knowledge Hypergraph RAG (OKH-RAG), which treats order as a first-class structural property. OKH-RAG represents knowledge as higher-order interactions within a hypergraph augmented with precedence structure, and reformulates retrieval as sequence inference over hyperedges. Instead of selecting independent facts, it recovers coherent interaction trajectories that reflect underlying reasoning processes. A learned transition model infers precedence directly from data without requiring explicit temporal supervision. We evaluate OKH-RAG on order-sensitive question answering and explanation tasks, including tropical cyclone and port operation scenarios. OKH-RAG consistently outperforms permutation-invariant baselines, and ablations show that these gains arise specifically from modeling interaction order. These results highlight a key limitation of set-based retrieval: effective reasoning requires not only retrieving relevant evidence, but organizing it into structured sequences.

评分理由: 论文研究大型视觉语言模型（VLMs）中的语义固化现象，属于大模型技术原理的创新研究。核心相关关键词：1) “Large Language Models” (8分)：论文明确研究VLMs，属于大语言模型范畴；2) “Post-training” (10分)：论文核心实验涉及后训练对齐，研究不同规则训练对模型性能的影响；3) “Instruction Tuning” (5分)：与提示干预和规则对齐相关；4) “Hallucination Mitigation” (5分)：研究模型固化和错误，与事实性/幻觉缓解相关；5) “Mechanistic Interpretability” (8分)：通过激活操控研究错误机制，属于可解释AI。其他关键词如MoE、量化、推理加速等未涉及。

大型视觉语言模型（VLMs）通常依赖于熟悉的语义先验，但现有评估方法未能清晰区分感知失败与规则映射失败。我们将这种行为定义为语义固着：即使提示指定了另一种同样有效的映射关系，模型仍会保持默认解释。为分离这一效应，我们提出了VLM-Fix基准测试，该测试基于四款抽象策略游戏构建，通过在成对的标准规则与逆向规则表述下评估相同的终局棋盘状态来实施控制。在14个开源与闭源VLM上的实验表明，模型在标准规则下的准确率始终优于逆向规则，揭示出显著的语义固着差距。提示干预实验验证了该机制：使用中性别名提示能大幅缩小逆向规则差距，而语义负载的别名提示则会重新扩大差距。模型的后训练过程表现出强烈的规则对齐特性：针对单一规则的训练能提升同规则迁移性能，但损害反规则迁移能力；而联合规则训练则能提升更广泛的迁移性能。为在合成游戏之外检验外部有效性，我们在VLMBias数据集上评估了类似的去熟悉化干预措施，并观察到相同的定性模式。最后，通过对模型深层激活向量进行定向干预，能够部分恢复性能下降，表明语义固着错误至少在深层表征中具备可编辑性。项目页面、代码及数据集详见https://maveryn.github.io/vlm-fix/。

Large vision-language models (VLMs) often rely on familiar semantic priors, but existing evaluations do not cleanly separate perception failures from rule-mapping failures. We study this behavior as semantic fixation: preserving a default interpretation even when the prompt specifies an alternative, equally valid mapping. To isolate this effect, we introduce VLM-Fix, a controlled benchmark over four abstract strategy games that evaluates identical terminal board states under paired standard and inverse rule formulations. Across 14 open and closed VLMs, accuracy consistently favors standard rules, revealing a robust semantic-fixation gap. Prompt interventions support this mechanism: neutral alias prompts substantially narrow the inverse-rule gap, while semantically loaded aliases reopen it. Post-training is strongly rule-aligned: training on one rule improves same-rule transfer but hurts opposite-rule transfer, while joint-rule training improves broader transfer. To test external validity beyond synthetic games, we evaluate analogous defamiliarization interventions on VLMBias and observe the same qualitative pattern. Finally, late-layer activation steering partially recovers degraded performance, indicating that semantic-fixation errors are at least partly editable in late representations. Project page, code, and dataset available at https://maveryn.github.io/vlm-fix/.

评分理由: 论文核心研究LLM生成多个候选答案后的最佳选择问题，提出Radial Consensus Score方法，因此与"Large Language Models"高度相关（10分）。论文涉及多智能体辩论和推理任务，与"LLM Agents"和"Multi-agent Systems"有一定关联（各5分）。方法旨在提高答案可靠性，与"Hallucination Mitigation"有一定关联（5分）。论文在长形式推理任务上测试，与"Chain of Thought"有一定关联（5分）。其他关键词如MoE、量化、训练方法等未在论文中涉及，得0分。

大型语言模型（LLM）针对给定提示频繁生成多个候选回答，但从中选择最可靠的答案仍具挑战性，尤其在正确性与表面多数一致发生偏离时。现有方法（如自洽性）依赖于离散投票，而基于概率的方法往往无法捕捉候选答案间的关联，或倾向于低估高质量但出现频率较低的回应，且未能充分利用答案表征的几何结构。为应对这些局限，我们提出径向共识分数（RCS），这是一种简单、高效且无需训练的最佳答案选择方法。RCS通过计算答案嵌入的加权弗雷歇均值（语义中心），并依据各候选答案到该中心的径向距离进行排序，从而建模语义共识。重要的是，RCS提供了一个通用框架，支持多种加权方案，包括均匀加权、基于频率和基于概率的变体，能够在完全适用于黑盒场景的同时，灵活整合一致信号与模型置信度。在涵盖短问答和长推理任务的七个基准测试及五个开源权重模型上的大量实验表明，RCS各变体均持续优于强基线方法，且随着采样预算增加，其优势更为显著。RCS还可作为多智能体辩论中多数投票的有效即插即用替代方案，并在黑盒场景中展现出强大的鲁棒性。总体而言，这些结果凸显了几何共识作为一种可扩展且广泛适用的原则，能够实现可靠的答案选择，其应用范围超越了多数投票，延伸至LLM推理中更具表达力与鲁棒性的聚合机制。

Large language models (LLMs) frequently generate multiple candidate responses for a given prompt, yet selecting the most reliable one remains challenging, especially when correctness diverges from surface-level majority agreement. Existing approaches, such as self-consistency, rely on discrete voting, while probability-based methods often fail to capture relationships among candidate answers or tend to underweight high-quality but less frequent responses, and do not fully leverage the geometric structure of answer representations. To address these limitations, we introduce Radial Consensus Score (RCS), a simple, efficient, and training-free method for best-of-N selection. RCS models semantic consensus by computing a weighted Fréchet mean (semantic center) of answer embeddings and ranking candidates by their radial distance to this center. Importantly, RCS provides a general framework that supports multiple weighting schemes, including uniform, frequency-based, and probability-based variants, enabling flexible integration of agreement signals and model confidence while remaining fully applicable in black-box settings. Extensive experiments across seven benchmarks covering short-form QA and long-form reasoning tasks, and five open-weight models, demonstrate that RCS variants consistently outperform strong baselines, with gains becoming more pronounced as the sampling budget increases. RCS also serves as an effective drop-in replacement for majority voting in multi-agent debate and exhibits strong robustness in black-box scenarios. Overall, these results highlight geometric consensus as a scalable and broadly applicable principle for reliable answer selection, extending beyond majority voting to more expressive and robust aggregation in LLM inference.

评分理由: 论文核心研究RAG系统在主观内容处理上的局限性，并提出Opinion-Aware RAG架构。与"Retrieval-Augmented Generation"高度相关（15分），因为全文围绕RAG系统展开。与"Large Language Models"相关（10分），因为论文使用LLM进行意见提取。与"Hallucination Mitigation"有一定关联（5分），因为论文讨论偏见和代表性不足问题，涉及事实性和透明度风险。其他关键词如MoE、SLMs、Scaling Laws、训练方法、推理优化、代理系统、模型压缩等均未在论文中涉及，得0分。

检索增强生成（RAG）系统已经改变了大型语言模型（LLM）获取外部知识的方式，但我们发现，当前实现方案存在偏向事实性、客观内容的偏差，这一点在现有优先考虑客观检索的基准测试和数据集上得到了印证。这种事实性偏差——将观点和多元视角视为噪声而非待综合的信息——限制了RAG系统在涉及主观内容的现实场景（从社交媒体讨论到产品评论）中的应用。除了技术局限，这种偏差还对透明和负责任的人工智能构成了风险：可能引发回音室效应而放大主流观点，导致少数群体声音系统性代表不足，以及通过有偏差的信息综合进行潜在观点操纵。我们通过不确定性视角将这一局限形式化：事实性查询涉及可通过证据减少的认知不确定性，而观点性查询则涉及反映人类视角真实异质性的偶然不确定性。这一区别意味着，事实性RAG应最小化后验熵，而观点感知型RAG必须保留它。基于这一理论基础，我们提出了一种观点感知型RAG架构，其特点包括基于LLM的观点提取、实体关联的观点图谱以及观点增强的文档索引。我们在电子商务卖家论坛数据上评估了该方法，将观点增强型知识库与传统基线进行对比。实验表明，检索多样性得到显著提升：在实体匹配文档上，情感多样性提升+26.8%，实体匹配率提升+42.7%，作者人口统计覆盖度提升+31.6%。我们的结果为“将主观性视为一等公民可产生可测量的、更具代表性的检索”提供了实证证据，这是迈向观点感知型RAG的第一步。未来工作包括为分布保真度而联合优化检索与生成。

RAG systems have transformed how LLMs access external knowledge, but we find that current implementations exhibit a bias toward factual, objective content, as evidenced by existing benchmarks and datasets that prioritize objective retrieval. This factual bias - treating opinions and diverse perspectives as noise rather than information to be synthesized - limits RAG systems in real-world scenarios involving subjective content, from social media discussions to product reviews. Beyond technical limitations, this bias poses risks to transparent and accountable AI: echo chamber effects that amplify dominant viewpoints, systematic underrepresentation of minority voices, and potential opinion manipulation through biased information synthesis. We formalize this limitation through the lens of uncertainty: factual queries involve epistemic uncertainty reducible through evidence, while opinion queries involve aleatoric uncertainty reflecting genuine heterogeneity in human perspectives. This distinction implies that factual RAG should minimize posterior entropy, whereas opinion-aware RAG must preserve it. Building on this theoretical foundation, we present an Opinion-Aware RAG architecture featuring LLM-based opinion extraction, entity-linked opinion graphs, and opinion-enriched document indexing. We evaluate our approach on e-commerce seller forum data, comparing an Opinion-Enriched knowledge base against a traditional baseline. Experiments demonstrate substantial improvements in retrieval diversity: +26.8% sentiment diversity, +42.7% entity match rate, and +31.6% author demographic coverage on entity-matched documents. Our results provide empirical evidence that treating subjectivity as a first-class citizen yields measurably more representative retrieval-a first step toward opinion-aware RAG. Future work includes joint optimization of retrieval and generation for distributional fidelity.

评分理由: 论文核心研究LLMs在道德决策中表现出的可识别受害者效应（IVE），并系统评估了16个前沿模型。高度相关的关键词包括：1）“Large Language Models”（论文研究对象）；2）“Instruction Tuning” OR “Alignment”（研究发现对齐训练强烈调节IVE，指令调优模型表现出极端IVE）；3）“Chain of Thought” OR “CoT Reasoning”（论文测试了标准CoT提示和功利主义CoT对IVE的影响）。其他关键词如MoE、SLMs、Scaling Laws、PEFT、RAG等均未在论文中涉及，故评分为0。

可识别受害者效应（Identifiable Victim Effect, IVE）——即人们倾向于向一个具体、通过叙事描述的受害者分配比向一个面临同等困境的统计特征群体分配更多资源的倾向——是道德心理学和行为经济学中最稳健的发现之一。随着大语言模型（LLMs）在人道主义分级诊疗、自动化资助评估和内容审核等领域承担起重要角色，一个关键问题随之产生：这些系统是否继承了人类道德推理中存在的情感非理性？我们首次对大语言模型中的IVE进行了系统性、大规模的实证研究，涵盖来自九个机构谱系（谷歌、Anthropic、OpenAI、Meta、DeepSeek、xAI、阿里巴巴、IBM和Moonshot）的16个前沿模型，共计N=51,955次经过验证的API试验。通过采用十项实验——移植并扩展了Small等人（2007）以及Kogut和Ritov（2005）的经典范式——我们发现IVE普遍存在，但受到对齐训练的强烈调节。经过指令微调的模型表现出极端的IVE（科恩d值高达1.56），而专长于推理的模型则逆转了该效应（d值低至-0.85）。汇总效应（d=0.223，p=2e-6）约为Lee和Feeley（2016）报告的单受害者人类元分析基线（d≈0.10）的两倍——并且很可能以更大的幅度超过人类的整体汇总效应，因为人类对群体受害者的效应近乎为零。标准的思维链（Chain-of-Thought, CoT）提示——与其作为审慎纠正机制的角色相反——几乎使IVE效应量增加了两倍（从d=0.15增至d=0.41），而只有功利主义的CoT提示能可靠地消除该效应。我们进一步记录了心理物理麻木、完全的数量忽视以及微小的内群体/外群体文化偏见，这些发现对于人工智能在人道主义和伦理决策情境中的部署具有重要启示。

The Identifiable Victim Effect (IVE) $-$ the tendency to allocate greater resources to a specific, narratively described victim than to a statistically characterized group facing equivalent hardship $-$ is one of the most robust findings in moral psychology and behavioural economics. As large language models (LLMs) assume consequential roles in humanitarian triage, automated grant evaluation, and content moderation, a critical question arises: do these systems inherit the affective irrationalities present in human moral reasoning? We present the first systematic, large-scale empirical investigation of the IVE in LLMs, comprising N=51,955 validated API trials across 16 frontier models spanning nine organizational lineages (Google, Anthropic, OpenAI, Meta, DeepSeek, xAI, Alibaba, IBM, and Moonshot). Using a suite of ten experiments $-$ porting and extending canonical paradigms from Small et al. (2007) and Kogut and Ritov (2005) $-$ we find that the IVE is prevalent but strongly modulated by alignment training. Instruction-tuned models exhibit extreme IVE (Cohen’s d up to 1.56), while reasoning-specialized models invert the effect (down to d=-0.85). The pooled effect (d=0.223, p=2e-6) is approximately twice the single-victim human meta-analytic baseline (d$\approx$0.10) reported by Lee and Feeley (2016) $-$ and likely exceeds the overall human pooled effect by a larger margin, given that the group-victim human effect is near zero. Standard Chain-of-Thought (CoT) prompting $-$ contrary to its role as a deliberative corrective $-$ nearly triples the IVE effect size (from d=0.15 to d=0.41), while only utilitarian CoT reliably eliminates it. We further document psychophysical numbing, perfect quantity neglect, and marginal in-group/out-group cultural bias, with implications for AI deployment in humanitarian and ethical decision-making contexts.

评分理由: 论文QuarkMedSearch专注于医疗领域的深度搜索智能体研究，基于基础模型Tongyi DeepResearch构建。核心相关关键词包括：1) “Large Language Models” (权重1.0) - 论文明确基于"agentic foundation models"，属于大模型应用，给10分；2) “Post-training” (权重1.0) - 论文采用两阶段SFT和RL训练策略，是核心方法，给10分；3) “RLHF” (权重1.0) - 论文提到RL训练，与RLHF相关但非核心，给5分；4) “Chain of Thought” (权重1.0) - 深度搜索涉及多步推理，给5分；5) “System 2 Thinking” (权重1.0) - 深度搜索需要深入推理，给5分；6) “Self-Correction” (权重1.0) - 论文强调反思能力，给10分；7) “LLM Agents” (权重1.0) - 论文是智能体研究，给10分；8) “Tool Use” (权重1.0) - 论文涉及工具调用，给10分；9) “AI for Science” (权重1.0) - 医疗AI应用，给10分。其他关键词如MoE、量化、RAG等未在摘要中体现，给0分。

随着具身化基础模型的持续演进，如何进一步提升其在垂直领域中的性能已成为重要挑战。为此，我们在强大的具身化基础模型通义DeepResearch基础上，聚焦中文医疗深度搜索场景，提出QuarkMedSearch，系统性地探索了涵盖医疗多跳数据构建、训练策略与评估基准的全流程方法，以进一步推进并评估其在垂直领域的性能上限。具体而言，在数据合成方面，针对医疗领域深度搜索训练数据稀缺的问题，我们结合大规模医疗知识图谱与实时在线探索，构建了长周期的医疗深度搜索训练数据；在训练后优化阶段，我们采用两阶段的监督微调（SFT）与强化学习（RL）训练策略，逐步增强模型进行深度搜索所需的规划、工具调用与反思能力，同时保持搜索效率；在评估方面，我们与医学专家合作，通过严格的人工验证构建了QuarkMedSearch Benchmark。实验结果表明，QuarkMedSearch在QuarkMedSearch Benchmark上取得了同规模开源模型中的最优性能，同时在通用基准测试中也保持了强劲的竞争力。

As agentic foundation models continue to evolve, how to further improve their performance in vertical domains has become an important challenge. To this end, building upon Tongyi DeepResearch, a powerful agentic foundation model, we focus on the Chinese medical deep search scenario and propose QuarkMedSearch, systematically exploring a full-pipeline approach spanning medical multi-hop data construction, training strategies, and evaluation benchmarks to further push and assess its performance upper bound in vertical domains. Specifically, for data synthesis, to address the scarcity of deep search training data in the medical domain, we combine a large-scale medical knowledge graph with real-time online exploration to construct long-horizon medical deep search training data; for post-training, we adopt a two-stage SFT and RL training strategy that progressively enhances the model’s planning, tool invocation, and reflection capabilities required for deep search, while maintaining search efficiency; for evaluation, we collaborate with medical experts to construct the QuarkMedSearch Benchmark through rigorous manual verification. Experimental results demonstrate that QuarkMedSearch achieves state-of-the-art performance among open-source models of comparable scale on the QuarkMedSearch Benchmark, while also maintaining strong competitiveness on general benchmarks.

评分理由: 论文核心研究LLM-based多模态情感识别，提出AffectAgent框架，包含三个协作智能体（query planner, evidence filter, emotion generator），使用Multi-Agent Proximal Policy Optimization (MAPPO)优化。高度相关的关键词包括：LLMs（核心基础）、MoE（提出MB-MoE方法）、RAG（检索增强生成框架）、LLM Agents和Multi-agent Systems（多智能体协作框架）。中等相关的关键词：Chain of Thought和System 2 Thinking（涉及分析推理过程）、Hallucination Mitigation（解决幻觉问题）。其他关键词如SLMs、Scaling Laws、Pre-training等未在论文中涉及。

基于大语言模型的多模态情感识别依赖静态参数化记忆，在解读细微情感状态时常产生幻觉。本文针对单轮检索增强生成方法易受模态模糊性影响、难以捕捉跨模态复杂情感依赖的问题，提出AffectAgent——一种面向情感的多智能体检索增强生成框架，该框架利用智能体间的协同决策实现细粒度情感理解。具体而言，AffectAgent包含三个联合优化的专用智能体：查询规划器、证据过滤器和情感生成器，它们通过协作式分析推理实现跨模态样本检索、证据评估与预测生成。这些智能体采用多智能体近端策略优化算法进行端到端优化，并通过共享情感奖励机制确保情感理解的一致性。此外，我们提出模态平衡专家混合模块与检索增强自适应融合模块：前者动态调节不同模态的贡献以缓解跨模态异质性导致的表征失配，后者通过引入检索到的视听嵌入向量增强缺失模态条件下的语义补全能力。在MER-UniBench基准上的大量实验表明，AffectAgent在复杂场景中均取得优越性能。代码发布于：https://github.com/Wz1h1NG/AffectAgent。

LLM-based multimodal emotion recognition relies on static parametric memory and often hallucinates when interpreting nuanced affective states. In this paper, given that single-round retrieval-augmented generation is highly susceptible to modal ambiguity and therefore struggles to capture complex affective dependencies across modalities, we introduce AffectAgent, an affect-oriented multi-agent retrieval-augmented generation framework that leverages collaborative decision-making among agents for fine-grained affective understanding. Specifically, AffectAgent comprises three jointly optimized specialized agents, namely a query planner, an evidence filter, and an emotion generator, which collaboratively perform analytical reasoning to retrieve cross-modal samples, assess evidence, and generate predictions. These agents are optimized end-to-end using Multi-Agent Proximal Policy Optimization (MAPPO) with a shared affective reward to ensure consistent emotion understanding. Furthermore, we introduce Modality-Balancing Mixture of Experts (MB-MoE) and Retrieval-Augmented Adaptive Fusion (RAAF), where MB-MoE dynamically regulates the contributions of different modalities to mitigate representation mismatch caused by cross-modal heterogeneity, while RAAF enhances semantic completion under missing-modality conditions by incorporating retrieved audiovisual embeddings. Extensive experiments on MER-UniBench demonstrate that AffectAgent achieves superior performance across complex scenarios. Our code will be released at: https://github.com/Wz1h1NG/AffectAgent.

评分理由: 论文提出LoSA方法，针对块状扩散语言模型（DLMs）在长上下文场景中的内存瓶颈问题，通过利用令牌隐藏状态的稳定性，设计了一种局部感知的稀疏注意力机制。该方法核心涉及稀疏模型（MoE/Sparse Models）和注意力优化技术（KV Cache Compression/Linear Attention/FlashAttention），以解决长上下文（Long Context LLMs）处理中的效率问题，并实现推理加速（Inference Acceleration）。论文虽未直接使用"LLMs"术语，但DLMs属于大语言模型的一种变体，因此给予中等相关度。其他关键词如小模型、训练方法、对齐、代理等与论文内容无关。

分块扩散语言模型（DLMs）能够以任意顺序生成多个词元，为自回归解码流程提供了一种有前景的替代方案。然而，在长上下文场景中，它们仍然受限于内存约束的注意力机制。由于存在KV膨胀问题，朴素的稀疏注意力在DLMs上效果不佳：不同的查询会选择不同的前缀位置，导致被访问的KV页面集合过大。为解决此问题，我们观察到在连续的降噪步骤之间，只有一小部分活跃词元的隐藏状态发生显著变化，而大多数稳定词元的状态几乎保持不变。基于这一洞察，我们提出了局部感知稀疏注意力（Locality-aware Sparse Attention, LOSA）。该方法对稳定词元复用已缓存的前缀注意力结果，仅对活跃词元应用稀疏注意力。这大幅减少了必须加载的KV索引数量，从而实现了更高的加速比和更高的准确性。在多种分块DLMs和基准测试中，LOSA在保持接近稠密模型准确性的同时，显著提升了效率：在激进的稀疏度水平下，平均准确率最高提升9个百分点，同时注意力密度降低至1.54倍以下。在RTX A6000 GPU上，注意力计算速度最高提升4.14倍，证明了所提方法的有效性。

Block-wise diffusion language models (DLMs) generate multiple tokens in any order, offering a promising alternative to the autoregressive decoding pipeline. However, they still remain bottlenecked by memory-bound attention in long-context scenarios. Naive sparse attention fails on DLMs due to a KV Inflation problem, where different queries select different prefix positions, making the union of accessed KV pages large. To address this, we observe that between consecutive denoising steps, only a small fraction of active tokens exhibit significant hidden-state changes, while the majority of stable tokens remain nearly constant. Based on this insight, we propose LOSA (Locality-aware Sparse Attention), which reuses cached prefix-attention results for stable tokens and applies sparse attention only to active tokens. This substantially shrinks the number of KV indices that must be loaded, yielding both higher speedup and higher accuracy. Across multiple block-wise DLMs and benchmarks, LOSA preserves near-dense accuracy while significantly improving efficiency, achieving up to +9 points in average accuracy at aggressive sparsity levels while maintaining 1.54x lower attention density. It also achieves up to 4.14x attention speedup on RTX A6000 GPUs, demonstrating the effectiveness of the proposed method.

评分理由: 论文核心研究LLMs处理长上下文时的效率瓶颈（KV缓存和计算复杂度），提出Latent-Condensed Attention（LCA）方法。高度相关（10分）的关键词：“Large Language Models”（论文明确针对LLMs）、“Context Window Extension”（解决长上下文建模）、“KV Cache Compression”（核心贡献是减少KV缓存）。中等相关（5分）的关键词：“Mixture of Experts”（提及稀疏方法作为背景）、“Quantization”（涉及模型效率优化）、“Speculative Decoding”（属于推理加速范畴）。其他关键词与论文内容无直接关联（0分）。

大语言模型在处理长上下文时面临显著挑战，这主要源于键值缓存（KV cache）的线性增长与自注意力机制（self-attention）的二次计算复杂度。现有方法通常分别应对这些瓶颈：多头潜在注意力（Multi-head Latent Attention, MLA）通过将词元投影至低维潜在空间以减少KV缓存，而稀疏注意力则致力于降低计算开销。然而，稀疏方法无法直接在MLA压缩后的潜在结构上运行，错失了联合优化的机会。本文提出潜在凝聚注意力（Latent-Condensed Attention, LCA），该方法直接在MLA的潜在空间内对上下文进行凝聚，其中表示被解耦为语义潜在向量与位置键。LCA通过查询感知池化分别聚合语义向量，并通过锚点选择保留位置键。这一设计在不增加参数的前提下，联合降低了计算成本与KV缓存。除MLA外，LCA的设计与架构无关，可轻松扩展至其他注意力机制，如分组查询注意力（GQA）。理论上，我们证明了其误差界与长度无关。实验表明，在128K上下文长度下，LCA实现了高达2.5倍的预填充加速与90%的KV缓存减少，同时保持了具有竞争力的性能。

Large language models (LLMs) face significant challenges in processing long contexts due to the linear growth of the key-value (KV) cache and quadratic complexity of self-attention. Existing approaches address these bottlenecks separately: Multi-head Latent Attention (MLA) reduces the KV cache by projecting tokens into a low-dimensional latent space, while sparse attention reduces computation. However, sparse methods cannot operate natively on MLA’s compressed latent structure, missing opportunities for joint optimization. In this paper, we propose Latent-Condensed Attention (LCA), which directly condenses context within MLA’s latent space, where the representation is disentangled into semantic latent vectors and positional keys. LCA separately aggregates semantic vectors via query-aware pooling and preserves positional keys via anchor selection. This approach jointly reduces both computational cost and KV cache without adding parameters. Beyond MLA, LCA’s design is architecture-agnostic and readily extends to other attention mechanisms such as GQA. Theoretically, we prove a length-independent error bound. Experiments show LCA achieves up to 2.5$\times$ prefilling speedup and 90% KV cache reduction at 128K context while maintaining competitive performance.

评分理由: 论文提出HypoExplore框架，使用大语言模型（LLM）作为核心组件来生成和评估神经网络架构假设，属于LLM在科学发现（AI for Science）领域的创新应用。框架本质上是基于LLM的自主智能体（LLM Agents）系统，通过多智能体（Multi-agent Systems）协作分析实验结果，并涉及假设驱动的推理过程（Chain of Thought/System 2 Thinking）和自我改进机制（Self-Correction）。论文主要聚焦于视觉架构发现，未涉及其他关键词如MoE、训练技术、推理优化等具体技术细节。

我们提出HypoExplore，这是一个智能体框架，它将视觉识别中的神经架构发现构建为一种假设驱动的科学探究过程。在给定人类指定的高层次研究方向后，HypoExplore通过演化分支进行神经架构的构思、实现、评估与改进。新假设的生成利用大型语言模型完成，通过选择父假设作为构建基础，并遵循一种双重策略进行引导：该策略既注重利用已验证的设计原则，也致力于解决尚不确定的问题。我们提出的框架维护着一个记录所有提出架构谱系的“轨迹树”，以及一个通过实验证据主动追踪置信度的“假设记忆库”。每次实验后，多个反馈智能体会从不同角度分析结果，并将其发现整合为假设置信度的更新。我们在CIFAR-10数据集上测试了该框架以发现轻量级视觉架构，其中最佳架构从初始基线精度18.91%演化至94.11%的准确率，并成功推广至CIFAR-100和Tiny-ImageNet。我们通过在MedMNIST数据集上进行独立的架构发现实验，进一步证明了该框架在专业领域的适用性，所获架构达到了最先进的性能。研究表明，随着证据的积累，假设置信度分数变得越来越具有预测性，并且学习到的设计原则能够在独立的演化谱系间迁移，这表明HypoExplore不仅能发现更强的架构，还有助于真正理解设计空间。

We introduce HypoExplore, an agentic framework that formulates neural architecture discovery for visual recognition as a hypothesis-driven scientific inquiry. Given a human-specified high-level research direction, HypoExplore ideates, implements, evaluates, and improves neural architectures through evolutionary branching. New hypotheses are created using a large language model by selecting a parent hypothesis to build upon, guided by a dual strategy that balances exploiting validated principles with resolving uncertain ones. Our proposed framework maintains a Trajectory Tree that records the lineage of all proposed architectures, and a Hypothesis Memory Bank that actively tracks confidence scores acquired through experimental evidence. After each experiment, multiple feedback agents analyze the results from different perspectives and consolidate their findings into hypothesis confidence updates. Our framework is tested on discovering lightweight vision architectures on CIFAR-10, with the best achieving 94.11% accuracy evolved from a root node baseline that starts at 18.91%, and generalizes to CIFAR-100 and Tiny-ImageNet. We further demonstrate applicability to a specialized domain by conducting independent architecture discovery runs on MedMNIST, which yield a state-of-the-art performance. We show that hypothesis confidence scores grow increasingly predictive as evidence accumulates, and that the learned principles transfer across independent evolutionary lineages, suggesting that HypoExplore not only discovers stronger architectures, but can help build a genuine understanding of the design space.

评分理由: 论文核心研究RAG方法创新，与"Retrieval-Augmented Generation"高度相关（15分），直接应用于大语言模型（10分）。论文关注顺序敏感推理，与"Chain of Thought"和"System 2 Thinking"有一定关联（各8分），涉及多步推理和深度推理过程。其他关键词如MoE、量化、对齐等未在论文中涉及，得0分。

检索增强生成（RAG）通过将输出建立在检索到的知识基础上，增强了大型语言模型的能力。然而，现有的RAG方法，包括基于图和超图的方法，都将检索到的证据视为无序集合，隐含地假设了排列不变性。这一假设与许多现实世界的推理任务不符，因为任务结果不仅取决于发生了哪些交互，还取决于这些交互展开的顺序。我们提出了顺序感知知识超图RAG（Order-Aware Knowledge Hypergraph RAG, OKH-RAG），它将顺序视为一种首要的结构属性。OKH-RAG将知识表示为带有优先序结构的超图中的高阶交互，并将检索重新定义为对超边的序列推断。它并非选择独立的事实，而是恢复反映底层推理过程的连贯交互轨迹。一个学习到的转移模型直接从数据中推断优先序，而无需显式的时间监督。我们在顺序敏感的问答和解释任务（包括热带气旋和港口运营场景）上评估了OKH-RAG。OKH-RAG始终优于排列不变的基线方法，消融实验表明，这些性能提升正是源于对交互顺序的建模。这些结果突显了基于集合的检索的一个关键局限：有效的推理不仅需要检索相关证据，还需要将其组织成结构化的序列。

Retrieval-augmented generation (RAG) enhances large language models by grounding outputs in retrieved knowledge. However, existing RAG methods including graph- and hypergraph-based approaches treat retrieved evidence as an unordered set, implicitly assuming permutation invariance. This assumption is misaligned with many real-world reasoning tasks, where outcomes depend not only on which interactions occur, but also on the order in which they unfold. We propose Order-Aware Knowledge Hypergraph RAG (OKH-RAG), which treats order as a first-class structural property. OKH-RAG represents knowledge as higher-order interactions within a hypergraph augmented with precedence structure, and reformulates retrieval as sequence inference over hyperedges. Instead of selecting independent facts, it recovers coherent interaction trajectories that reflect underlying reasoning processes. A learned transition model infers precedence directly from data without requiring explicit temporal supervision. We evaluate OKH-RAG on order-sensitive question answering and explanation tasks, including tropical cyclone and port operation scenarios. OKH-RAG consistently outperforms permutation-invariant baselines, and ablations show that these gains arise specifically from modeling interaction order. These results highlight a key limitation of set-based retrieval: effective reasoning requires not only retrieving relevant evidence, but organizing it into structured sequences.

评分理由: 论文研究大型视觉语言模型（VLMs）中的语义固化现象，属于大模型技术原理的创新研究。核心相关关键词：1) “Large Language Models” (8分)：论文明确研究VLMs，属于大语言模型范畴；2) “Post-training” (10分)：论文核心实验涉及后训练对齐，研究不同规则训练对模型性能的影响；3) “Instruction Tuning” (5分)：与提示干预和规则对齐相关；4) “Hallucination Mitigation” (5分)：研究模型固化和错误，与事实性/幻觉缓解相关；5) “Mechanistic Interpretability” (8分)：通过激活操控研究错误机制，属于可解释AI。其他关键词如MoE、量化、推理加速等未涉及。

大型视觉语言模型（VLMs）通常依赖于熟悉的语义先验，但现有评估方法未能清晰区分感知失败与规则映射失败。我们将这种行为定义为语义固着：即使提示指定了另一种同样有效的映射关系，模型仍会保持默认解释。为分离这一效应，我们提出了VLM-Fix基准测试，该测试基于四款抽象策略游戏构建，通过在成对的标准规则与逆向规则表述下评估相同的终局棋盘状态来实施控制。在14个开源与闭源VLM上的实验表明，模型在标准规则下的准确率始终优于逆向规则，揭示出显著的语义固着差距。提示干预实验验证了该机制：使用中性别名提示能大幅缩小逆向规则差距，而语义负载的别名提示则会重新扩大差距。模型的后训练过程表现出强烈的规则对齐特性：针对单一规则的训练能提升同规则迁移性能，但损害反规则迁移能力；而联合规则训练则能提升更广泛的迁移性能。为在合成游戏之外检验外部有效性，我们在VLMBias数据集上评估了类似的去熟悉化干预措施，并观察到相同的定性模式。最后，通过对模型深层激活向量进行定向干预，能够部分恢复性能下降，表明语义固着错误至少在深层表征中具备可编辑性。项目页面、代码及数据集详见https://maveryn.github.io/vlm-fix/。

Large vision-language models (VLMs) often rely on familiar semantic priors, but existing evaluations do not cleanly separate perception failures from rule-mapping failures. We study this behavior as semantic fixation: preserving a default interpretation even when the prompt specifies an alternative, equally valid mapping. To isolate this effect, we introduce VLM-Fix, a controlled benchmark over four abstract strategy games that evaluates identical terminal board states under paired standard and inverse rule formulations. Across 14 open and closed VLMs, accuracy consistently favors standard rules, revealing a robust semantic-fixation gap. Prompt interventions support this mechanism: neutral alias prompts substantially narrow the inverse-rule gap, while semantically loaded aliases reopen it. Post-training is strongly rule-aligned: training on one rule improves same-rule transfer but hurts opposite-rule transfer, while joint-rule training improves broader transfer. To test external validity beyond synthetic games, we evaluate analogous defamiliarization interventions on VLMBias and observe the same qualitative pattern. Finally, late-layer activation steering partially recovers degraded performance, indicating that semantic-fixation errors are at least partly editable in late representations. Project page, code, and dataset available at https://maveryn.github.io/vlm-fix/.

评分理由: 论文核心研究LLM生成多个候选答案后的最佳选择问题，提出Radial Consensus Score方法，因此与"Large Language Models"高度相关（10分）。论文涉及多智能体辩论和推理任务，与"LLM Agents"和"Multi-agent Systems"有一定关联（各5分）。方法旨在提高答案可靠性，与"Hallucination Mitigation"有一定关联（5分）。论文在长形式推理任务上测试，与"Chain of Thought"有一定关联（5分）。其他关键词如MoE、量化、训练方法等未在论文中涉及，得0分。

大型语言模型（LLM）针对给定提示频繁生成多个候选回答，但从中选择最可靠的答案仍具挑战性，尤其在正确性与表面多数一致发生偏离时。现有方法（如自洽性）依赖于离散投票，而基于概率的方法往往无法捕捉候选答案间的关联，或倾向于低估高质量但出现频率较低的回应，且未能充分利用答案表征的几何结构。为应对这些局限，我们提出径向共识分数（RCS），这是一种简单、高效且无需训练的最佳答案选择方法。RCS通过计算答案嵌入的加权弗雷歇均值（语义中心），并依据各候选答案到该中心的径向距离进行排序，从而建模语义共识。重要的是，RCS提供了一个通用框架，支持多种加权方案，包括均匀加权、基于频率和基于概率的变体，能够在完全适用于黑盒场景的同时，灵活整合一致信号与模型置信度。在涵盖短问答和长推理任务的七个基准测试及五个开源权重模型上的大量实验表明，RCS各变体均持续优于强基线方法，且随着采样预算增加，其优势更为显著。RCS还可作为多智能体辩论中多数投票的有效即插即用替代方案，并在黑盒场景中展现出强大的鲁棒性。总体而言，这些结果凸显了几何共识作为一种可扩展且广泛适用的原则，能够实现可靠的答案选择，其应用范围超越了多数投票，延伸至LLM推理中更具表达力与鲁棒性的聚合机制。

Large language models (LLMs) frequently generate multiple candidate responses for a given prompt, yet selecting the most reliable one remains challenging, especially when correctness diverges from surface-level majority agreement. Existing approaches, such as self-consistency, rely on discrete voting, while probability-based methods often fail to capture relationships among candidate answers or tend to underweight high-quality but less frequent responses, and do not fully leverage the geometric structure of answer representations. To address these limitations, we introduce Radial Consensus Score (RCS), a simple, efficient, and training-free method for best-of-N selection. RCS models semantic consensus by computing a weighted Fréchet mean (semantic center) of answer embeddings and ranking candidates by their radial distance to this center. Importantly, RCS provides a general framework that supports multiple weighting schemes, including uniform, frequency-based, and probability-based variants, enabling flexible integration of agreement signals and model confidence while remaining fully applicable in black-box settings. Extensive experiments across seven benchmarks covering short-form QA and long-form reasoning tasks, and five open-weight models, demonstrate that RCS variants consistently outperform strong baselines, with gains becoming more pronounced as the sampling budget increases. RCS also serves as an effective drop-in replacement for majority voting in multi-agent debate and exhibits strong robustness in black-box scenarios. Overall, these results highlight geometric consensus as a scalable and broadly applicable principle for reliable answer selection, extending beyond majority voting to more expressive and robust aggregation in LLM inference.

评分理由: 论文核心研究RAG系统在主观内容处理上的局限性，并提出Opinion-Aware RAG架构。与"Retrieval-Augmented Generation"高度相关（15分），因为全文围绕RAG系统展开。与"Large Language Models"相关（10分），因为论文使用LLM进行意见提取。与"Hallucination Mitigation"有一定关联（5分），因为论文讨论偏见和代表性不足问题，涉及事实性和透明度风险。其他关键词如MoE、SLMs、Scaling Laws、训练方法、推理优化、代理系统、模型压缩等均未在论文中涉及，得0分。

检索增强生成（RAG）系统已经改变了大型语言模型（LLM）获取外部知识的方式，但我们发现，当前实现方案存在偏向事实性、客观内容的偏差，这一点在现有优先考虑客观检索的基准测试和数据集上得到了印证。这种事实性偏差——将观点和多元视角视为噪声而非待综合的信息——限制了RAG系统在涉及主观内容的现实场景（从社交媒体讨论到产品评论）中的应用。除了技术局限，这种偏差还对透明和负责任的人工智能构成了风险：可能引发回音室效应而放大主流观点，导致少数群体声音系统性代表不足，以及通过有偏差的信息综合进行潜在观点操纵。我们通过不确定性视角将这一局限形式化：事实性查询涉及可通过证据减少的认知不确定性，而观点性查询则涉及反映人类视角真实异质性的偶然不确定性。这一区别意味着，事实性RAG应最小化后验熵，而观点感知型RAG必须保留它。基于这一理论基础，我们提出了一种观点感知型RAG架构，其特点包括基于LLM的观点提取、实体关联的观点图谱以及观点增强的文档索引。我们在电子商务卖家论坛数据上评估了该方法，将观点增强型知识库与传统基线进行对比。实验表明，检索多样性得到显著提升：在实体匹配文档上，情感多样性提升+26.8%，实体匹配率提升+42.7%，作者人口统计覆盖度提升+31.6%。我们的结果为“将主观性视为一等公民可产生可测量的、更具代表性的检索”提供了实证证据，这是迈向观点感知型RAG的第一步。未来工作包括为分布保真度而联合优化检索与生成。

RAG systems have transformed how LLMs access external knowledge, but we find that current implementations exhibit a bias toward factual, objective content, as evidenced by existing benchmarks and datasets that prioritize objective retrieval. This factual bias - treating opinions and diverse perspectives as noise rather than information to be synthesized - limits RAG systems in real-world scenarios involving subjective content, from social media discussions to product reviews. Beyond technical limitations, this bias poses risks to transparent and accountable AI: echo chamber effects that amplify dominant viewpoints, systematic underrepresentation of minority voices, and potential opinion manipulation through biased information synthesis. We formalize this limitation through the lens of uncertainty: factual queries involve epistemic uncertainty reducible through evidence, while opinion queries involve aleatoric uncertainty reflecting genuine heterogeneity in human perspectives. This distinction implies that factual RAG should minimize posterior entropy, whereas opinion-aware RAG must preserve it. Building on this theoretical foundation, we present an Opinion-Aware RAG architecture featuring LLM-based opinion extraction, entity-linked opinion graphs, and opinion-enriched document indexing. We evaluate our approach on e-commerce seller forum data, comparing an Opinion-Enriched knowledge base against a traditional baseline. Experiments demonstrate substantial improvements in retrieval diversity: +26.8% sentiment diversity, +42.7% entity match rate, and +31.6% author demographic coverage on entity-matched documents. Our results provide empirical evidence that treating subjectivity as a first-class citizen yields measurably more representative retrieval-a first step toward opinion-aware RAG. Future work includes joint optimization of retrieval and generation for distributional fidelity.

评分理由: 论文核心研究LLMs在道德决策中表现出的可识别受害者效应（IVE），并系统评估了16个前沿模型。高度相关的关键词包括：1）“Large Language Models”（论文研究对象）；2）“Instruction Tuning” OR “Alignment”（研究发现对齐训练强烈调节IVE，指令调优模型表现出极端IVE）；3）“Chain of Thought” OR “CoT Reasoning”（论文测试了标准CoT提示和功利主义CoT对IVE的影响）。其他关键词如MoE、SLMs、Scaling Laws、PEFT、RAG等均未在论文中涉及，故评分为0。

可识别受害者效应（Identifiable Victim Effect, IVE）——即人们倾向于向一个具体、通过叙事描述的受害者分配比向一个面临同等困境的统计特征群体分配更多资源的倾向——是道德心理学和行为经济学中最稳健的发现之一。随着大语言模型（LLMs）在人道主义分级诊疗、自动化资助评估和内容审核等领域承担起重要角色，一个关键问题随之产生：这些系统是否继承了人类道德推理中存在的情感非理性？我们首次对大语言模型中的IVE进行了系统性、大规模的实证研究，涵盖来自九个机构谱系（谷歌、Anthropic、OpenAI、Meta、DeepSeek、xAI、阿里巴巴、IBM和Moonshot）的16个前沿模型，共计N=51,955次经过验证的API试验。通过采用十项实验——移植并扩展了Small等人（2007）以及Kogut和Ritov（2005）的经典范式——我们发现IVE普遍存在，但受到对齐训练的强烈调节。经过指令微调的模型表现出极端的IVE（科恩d值高达1.56），而专长于推理的模型则逆转了该效应（d值低至-0.85）。汇总效应（d=0.223，p=2e-6）约为Lee和Feeley（2016）报告的单受害者人类元分析基线（d≈0.10）的两倍——并且很可能以更大的幅度超过人类的整体汇总效应，因为人类对群体受害者的效应近乎为零。标准的思维链（Chain-of-Thought, CoT）提示——与其作为审慎纠正机制的角色相反——几乎使IVE效应量增加了两倍（从d=0.15增至d=0.41），而只有功利主义的CoT提示能可靠地消除该效应。我们进一步记录了心理物理麻木、完全的数量忽视以及微小的内群体/外群体文化偏见，这些发现对于人工智能在人道主义和伦理决策情境中的部署具有重要启示。

The Identifiable Victim Effect (IVE) $-$ the tendency to allocate greater resources to a specific, narratively described victim than to a statistically characterized group facing equivalent hardship $-$ is one of the most robust findings in moral psychology and behavioural economics. As large language models (LLMs) assume consequential roles in humanitarian triage, automated grant evaluation, and content moderation, a critical question arises: do these systems inherit the affective irrationalities present in human moral reasoning? We present the first systematic, large-scale empirical investigation of the IVE in LLMs, comprising N=51,955 validated API trials across 16 frontier models spanning nine organizational lineages (Google, Anthropic, OpenAI, Meta, DeepSeek, xAI, Alibaba, IBM, and Moonshot). Using a suite of ten experiments $-$ porting and extending canonical paradigms from Small et al. (2007) and Kogut and Ritov (2005) $-$ we find that the IVE is prevalent but strongly modulated by alignment training. Instruction-tuned models exhibit extreme IVE (Cohen’s d up to 1.56), while reasoning-specialized models invert the effect (down to d=-0.85). The pooled effect (d=0.223, p=2e-6) is approximately twice the single-victim human meta-analytic baseline (d$\approx$0.10) reported by Lee and Feeley (2016) $-$ and likely exceeds the overall human pooled effect by a larger margin, given that the group-victim human effect is near zero. Standard Chain-of-Thought (CoT) prompting $-$ contrary to its role as a deliberative corrective $-$ nearly triples the IVE effect size (from d=0.15 to d=0.41), while only utilitarian CoT reliably eliminates it. We further document psychophysical numbing, perfect quantity neglect, and marginal in-group/out-group cultural bias, with implications for AI deployment in humanitarian and ethical decision-making contexts.

评分理由: 论文研究变分语言模型中的代理控制，核心是使用不确定性作为主动控制信号来调节训练、保留检查点和指导推理干预，实现闭环内部控制。与"LLM Agents"高度相关（10分），因为论文明确研究代理控制、代理评估和代理路由。与"Large Language Models"相关（8分），因为研究变分语言模型作为基础。与"Self-Correction"相关（8分），因为涉及自我调节、自我改进的控制机制。其他关键词如MoE、SFT、RAG等未在论文中涉及，得0分。

本研究探讨变分语言模型是否能够基于其内部证据支持一种最小化且可测量的主体控制形式。我们的模型结合了局部变分隐式计算（EVE）、稳态潜在调节器、结构感知的检查点保留机制，以及在保留模型之上运行的校准化不确定性感知控制器。我们并非将不确定性视为预测后被动测量的诊断指标，而是将其作为一种可调节训练、支持检查点保留并指导推理时干预的操作性信号。由此构建的框架具有明确的聚焦性：它研究一种闭环式的内部控制形式，使结构性与预测性信号可转化为具体操作。实证结果表明，该变分主干模型在语言建模任务上优于匹配的确定性参照模型，同时展现出更丰富且更可用的不确定性特征。在此主干模型基础上，校准控制器持续保持活跃状态，在完整的主体评估中使用多种操作，并实现了质量与成本的正向权衡。这些结果支持一个精确论断：内部不确定性不仅可以作为变分语言模型的描述性属性，更可作为调节、检查点保留及最小化主体路由的实用控制接口。

We study whether a variational language model can support a minimal and measurable form of agentic control grounded in its own internal evidence. Our model combines local variational hidden computation (EVE), a homeostatic latent regulator, structurally aware checkpoint retention and a calibrated uncertainty-aware controller operating on top of the retained model. Rather than treating uncertainty as a passive diagnostic measured after prediction, we treat it as an operational signal that can regulate training, support checkpoint retention and guide inference-time intervention. The resulting framework is deliberately focused. It studies a closed-loop form of internal control in which structural and predictive signals become actionable. Empirically, the variational backbone improves over a matched deterministic reference on the language-modeling task while also exhibiting a richer and more usable uncertainty profile. On top of this backbone, the calibrated controller remains active, uses multiple actions under a full agentic evaluation and yields a positive quality-cost trade-off. These results support a precise claim: internal uncertainty can serve not only as a descriptive property of a variational language model, but also as a practical control interface for regulation, checkpoint retention and minimal agentic routing.

评分理由: 论文核心研究LLM agents在模拟公民对官僚主义的情感反应方面的应用，并评估其与文化对齐的能力。因此，与"Large Language Models"和"LLM Agents"高度相关（10分）。论文涉及评估模型与人类情感反应的"alignment"，因此与"Instruction Tuning" OR “Alignment"有一定关联（5分）。其他关键词如MoE、SLMs、Scaling Laws、Pre-training、RLHF、RAG、Context Window、KV Cache、Reasoning、Self-Correction、Tool Use、Multi-agent、Quantization、Speculative Decoding、Hallucination、Interpretability、World Models、Model Merging、In-context Learning、AI for Science等均未在论文中提及或相关，故给0分。

提升政策制定效能是公共行政领域的核心关切。既有的人类主体研究表明，在政策执行过程中，公民对繁文缛节的情绪反应存在显著的跨文化差异。尽管大型语言模型（LLM）智能体为模拟类人反应和降低实验成本提供了机遇，但其能否针对繁文缛节产生符合文化背景的情绪反应仍未得到验证。为填补这一空白，我们提出了一个评估框架，用以评估LLM在不同文化背景下对繁文缛节的情绪反应。作为一项试点研究，我们将此框架应用于一个单一的繁文缛节场景。结果显示，所有模型与人类情绪反应的契合度均有限，在东方文化背景下的表现尤为薄弱。文化提示策略在提升契合度方面基本无效。我们进一步介绍了 \textbf{RAMO}，这是一个用于模拟公民对繁文缛节情绪反应并收集人类数据以改进模型的交互式界面。该界面已公开于 https://ramo-chi.ivia.ch。

Improving policymaking is a central concern in public administration. Prior human subject studies reveal substantial cross-cultural differences in citizens’ emotional responses to red tape during policy implementation. While LLM agents offer opportunities to simulate human-like responses and reduce experimental costs, their ability to generate culturally appropriate emotional responses to red tape remains unverified. To address this gap, we propose an evaluation framework for assessing LLMs’ emotional responses to red tape across diverse cultural contexts. As a pilot study, we apply this framework to a single red-tape scenario. Our results show that all models exhibit limited alignment with human emotional responses, with notably weaker performance in Eastern cultures. Cultural prompting strategies prove largely ineffective in improving alignment. We further introduce \textbf{RAMO}, an interactive interface for simulating citizens’ emotional responses to red tape and for collecting human data to improve models. The interface is publicly available at https://ramo-chi.ivia.ch.

评分理由: 论文核心研究LLMs与人类在System 1/System 2思维模式下的语义记忆结构与隐性偏见的关系，直接涉及"Large Language Models"和"System 2 Thinking"关键词，因此这两项给10分。论文通过比较人类与LLMs的认知机制来探索偏见根源，与"Mechanistic Interpretability"有一定关联，给5分。其他关键词如MoE、SFT、RAG等涉及具体技术方法，论文未涉及，给0分。

人类与大型语言模型（LLM）中存在的隐性偏见构成了显著的社会风险。双重加工理论认为，偏见主要源于联想式的系统1思维，而审慎的系统2思维则能减轻偏见，但导致这一现象的认知机制仍不甚明晰。为更好地理解人类（以及可能的LLM）中这种双重性的本质，我们将系统1和系统2思维建模为具有不同结构的语义记忆网络，这些网络基于人类和LLM生成的同类数据集构建。随后，我们运用基于网络的评估指标，探究这些不同的语义记忆结构如何与隐性性别偏见相关联。研究发现，语义记忆结构的不可约简性仅存在于人类中，这表明LLM缺乏某些类人化的概念知识。此外，仅在人类中，语义记忆结构与隐性偏见始终存在关联，且系统2结构中的偏见水平较低。这些发现表明，特定类型的概念知识有助于人类进行偏见调控，但对LLM无效，从而凸显了人类认知与机器认知之间的根本差异。

Implicit biases in both humans and large language models (LLMs) pose significant societal risks. Dual process theories propose that biases arise primarily from associative System 1 thinking, while deliberative System 2 thinking mitigates bias, but the cognitive mechanisms that give rise to this phenomenon remain poorly understood. To better understand what underlies this duality in humans, and possibly in LLMs, we model System 1 and System 2 thinking as semantic memory networks with distinct structures, built from comparable datasets generated by both humans and LLMs. We then investigate how these distinct semantic memory structures relate to implicit gender bias using network-based evaluation metrics. We find that semantic memory structures are irreducible only in humans, suggesting that LLMs lack certain types of human-like conceptual knowledge. Moreover, semantic memory structure relates consistently to implicit bias only in humans, with lower levels of bias in System~2 structures. These findings suggest that certain types of conceptual knowledge contribute to bias regulation in humans, but not in LLMs, highlighting fundamental differences between human and machine cognition.

评分理由: 论文TriFit专注于蛋白质突变适应性预测，属于生物信息学领域，与"AI for Science"高度相关（10分）。它使用ESM-2（一种蛋白质语言模型）提取序列嵌入，因此与"Large Language Models"有一定关联（5分）。核心创新是四专家混合专家（MoE）融合模块，与"Mixture of Experts"高度相关（10分）。其他关键词如SLMs、Scaling Laws、训练方法、推理优化、代理系统等均未涉及，得0分。

预测单氨基酸取代（SAVs）的功能影响是理解遗传疾病和设计治疗性蛋白质的核心。虽然蛋白质语言模型和基于结构的方法在此任务上已取得优异表现，但它们系统性地忽略了蛋白质动力学；残基柔性、协同运动及变构耦合是结构生物学中公认的突变耐受性决定因素，却尚未被纳入监督式变异效应预测器中。本文提出TriFit——一个多模态框架，它通过四专家混合专家（MoE）融合模块结合三模态跨模态对比学习，整合了序列、结构与蛋白质动力学信息。序列嵌入通过ESM-2（650M）的掩蔽边际评分提取；结构嵌入来自AlphaFold2预测的C-alpha几何构型；动力学嵌入则源自高斯网络模型（GNM）的B因子、模态振型及残基-残基互相关数据。MoE路由器根据输入自适应加权模态组合，实现了无需固定模态假设的蛋白质特异性融合。在ProteinGym取代基准测试（217个DMS实验，69.6万个SAVs）中，TriFit取得了AUROC 0.897 +/- 0.0002，超越了所有监督基线模型（包括Kermut的0.864和ProteinNPT的0.844）以及最佳零样本模型ESM3（0.769）。消融实验证实，动力学特征在成对模态组合基础上提供了最大的边际贡献，且TriFit无需后验校正即可获得校准良好的概率输出（ECE = 0.044）。

Predicting the functional impact of single amino acid substitutions (SAVs) is central to understanding genetic disease and engineering therapeutic proteins. While protein language models and structure-based methods have achieved strong performance on this task, they systematically neglect protein dynamics; residue flexibility, correlated motions, and allosteric coupling are well-established determinants of mutational tolerance in structural biology, yet have not been incorporated into supervised variant effect predictors. We present TriFit, a multimodal framework that integrates sequence, structure, and protein dynamics through a four-expert Mixture-of-Experts (MoE) fusion module with trimodal cross-modal contrastive learning. Sequence embeddings are extracted via masked marginal scoring with ESM-2 (650M); structural embeddings from AlphaFold2-predicted C-alpha geometries; and dynamics embeddings from Gaussian Network Model (GNM) B-factors, mode shapes, and residue-residue cross-correlations. The MoE router adaptively weights modality combinations conditioned on the input, enabling protein-specific fusion without fixed modality assumptions. On the ProteinGym substitution benchmark (217 DMS assays, 696k SAVs), TriFit achieves AUROC 0.897 +/- 0.0002, outperforming all supervised baselines including Kermut (0.864) and ProteinNPT (0.844), and the best zero-shot model ESM3 (0.769). Ablation studies confirm that dynamics provides the largest marginal contribution over pairwise modality combinations, and TriFit achieves well-calibrated probabilistic outputs (ECE = 0.044) without post-hoc correction.

评分理由: 论文核心研究LLMs在政治敏感环境中的隐私合规部署，提出了一种顺序遗忘框架，直接涉及LLMs和微调技术。与"Large Language Models"和"Post-training"高度相关（10分），因为论文明确研究LLMs并采用正负微调方法。其他关键词如MoE、SLMs、Scaling Laws、RAG、Agents等均未在标题或摘要中提及，与论文内容无关（0分）。

大型语言模型正日益部署于政治敏感环境中，其可能记忆个人数据或机密内容的特性引发了GDPR等监管框架下“被遗忘权”相关的合规担忧。将此类法律原则转化为适用于大规模生成式系统的技术方案，面临着显著挑战。

本文提出一种轻量级顺序遗忘框架，该框架明确区分知识保留与信息抑制目标。该方法首先通过正向微调稳定模型的良性能力，随后实施分层受限的负向微调，在保持通用语言能力的同时抑制指定的敏感信息模式。

在SemEval-2025 LLM遗忘基准测试中的实验表明，该方法能有效实现行为抑制，且对事实准确性与语言流畅性影响最小。GPT-2相比DistilGPT-2展现出更强的鲁棒性，这揭示了模型容量在隐私对齐适应中的关键作用。我们将顺序遗忘定位为一种可操作、可复现的机制，能够为政治场景中部署的LLM实现数据擦除需求提供实践路径。

Large Language Models (LLMs) are increasingly deployed in politically sensitive environments, where memorisation of personal data or confidential content raises regulatory concerns under frameworks such as the GDPR and its Right to be Forgotten. Translating such legal principles into large-scale generative systems presents significant technical challenges. We introduce a lightweight sequential unlearning framework that explicitly separates retention and suppression objectives. The method first stabilises benign capabilities through positive fine-tuning, then applies layer-restricted negative fine-tuning to suppress designated sensitive patterns while preserving general language competence. Experiments on the SemEval-2025 LLM Unlearning benchmark demonstrate effective behavioural suppression with minimal impact on factual accuracy and fluency. GPT-2 exhibits greater robustness than DistilGPT-2, highlighting the role of model capacity in privacy-aligned adaptation. We position sequential unlearning as a practical and reproducible mechanism for operationalising data erasure requirements in politically deployed LLMs.

评分理由: 论文提出了一种名为Chain-of-Models Pre-Training (CoM-PT)的新方法，旨在加速视觉基础模型（VFMs）的训练。该方法的核心是构建一个按模型大小升序排列的模型链，通过从较小的模型向较大的模型进行顺序逆向知识迁移，实现训练加速。论文与关键词的相关性分析如下：1）与"Pre-training"高度相关（10分），因为该方法专注于预训练阶段的加速，是论文的核心内容。2）与"Large Language Models” OR “LLMs” OR “Foundation Models"有一定关联（8分），论文虽然主要针对视觉基础模型，但摘要中提到该方法可扩展至大语言模型预训练，且"Foundation Models"一词与论文主题直接相关。3）其他关键词（如MoE、SFT、RAG、Quantization等）与论文内容无直接关联，因此评分为0分。

本文提出链式模型预训练（Chain-of-Models Pre-Training, CoM-PT），一种面向视觉基础模型（Vision Foundation Models, VFMs）的新型无损性能训练加速方法。该方法的核心动机与现有加速方法存在根本差异：其目标并非优化单个模型，而是在模型家族层面加速整个训练流程，并能随模型家族扩展高效扩展。具体而言，CoM-PT为模型家族建立按模型规模升序排列的预训练序列，称为模型链。在该链中，仅最小模型需进行标准的独立预训练，其余模型则通过联合复用参数空间与特征空间中的知识，从其较小的前驱模型进行序列化逆向知识迁移，从而实现高效训练。因此，CoM-PT使所有模型在显著降低训练成本的同时，获得普遍优于标准独立训练的性能，这一结论在涵盖零样本与微调任务的45个数据集上得到广泛验证。值得注意的是，其高效扩展特性带来一个显著现象：训练更多模型甚至能带来更高的效率。例如，在CC3M数据集上进行预训练时：i）以ViT-L作为最大模型，在模型链中逐步前置更小的模型可将计算复杂度降低高达72%；ii）在固定模型规模范围内，当VFM家族扩展至3、4、7个模型时，CoM-PT的加速比呈现出显著跃升：从4.13倍提升至5.68倍及7.09倍。由于CoM-PT天然与具体预训练范式无关，我们开源代码以促进其在计算密集型场景（如大语言模型预训练）中的进一步扩展。

In this paper, we present Chain-of-Models Pre-Training (CoM-PT), a novel performance-lossless training acceleration method for vision foundation models (VFMs). This approach fundamentally differs from existing acceleration methods in its core motivation: rather than optimizing each model individually, CoM-PT is designed to accelerate the training pipeline at the model family level, scaling efficiently as the model family expands. Specifically, CoM-PT establishes a pre-training sequence for the model family, arranged in ascending order of model size, called model chain. In this chain, only the smallest model undergoes standard individual pre-training, while the other models are efficiently trained through sequential inverse knowledge transfer from their smaller predecessors by jointly reusing the knowledge in the parameter space and the feature space. As a result, CoM-PT enables all models to achieve performance that is mostly superior to standard individual training while significantly reducing training cost, and this is extensively validated across 45 datasets spanning zero-shot and fine-tuning tasks. Notably, its efficient scaling property yields a remarkable phenomenon: training more models even results in higher efficiency. For instance, when pre-training on CC3M: i) given ViT-L as the largest model, progressively prepending smaller models to the model chain reduces computational complexity by up to 72%; ii) within a fixed model size range, as the VFM family scales across 3, 4, and 7 models, the acceleration ratio of CoM-PT exhibits a striking leap: from 4.13X to 5.68X and 7.09X. Since CoM-PT is naturally agnostic to specific pre-training paradigms, we open-source the code to spur further extensions in more computationally intensive scenarios, such as large language model pre-training.

评分理由: 论文主要研究中文毒性检测的可解释性方法，使用BERT风格编码器，并提到使用轻量级LLM指导来细化显著性线索。因此，与"Large Language Models"有一定关联（5分），因为使用了LLM指导；与"Mechanistic Interpretability"高度相关（10分），因为论文核心是可解释性AI方法。其他关键词如MoE、SLMs、Scaling Laws、Pre-training、RLHF、RAG、Agents、Quantization等均未在摘要中提及或相关，故评0分。

现有中文有害内容检测方法主要针对句子级分类，但往往无法提供可读且连贯的有害证据片段。我们提出ToxiTrace，一种面向可解释性的BERT风格编码器方法，包含三个核心组件：（1）CuSA，在轻量化大语言模型引导下将编码器衍生的显著性线索细化为细粒度有害文本片段；（2）GCLoss，一种梯度约束目标函数，可将词元级显著性聚焦于有害证据同时抑制无关激活；（3）ARCL，通过构建样本特异性对比推理对来锐化有害与非有害内容之间的语义边界。实验表明，ToxiTrace在保持高效编码器推理能力的同时，提升了分类准确率与有害片段提取性能，并能生成更连贯、可人工解读的解释。模型已发布于https://huggingface.co/ArdLi/ToxiTrace。

Existing Chinese toxic content detection methods mainly target sentence-level classification but often fail to provide readable and contiguous toxic evidence spans. We propose \textbf{ToxiTrace}, an explainability-oriented method for BERT-style encoders with three components: (1) \textbf{CuSA}, which refines encoder-derived saliency cues into fine-grained toxic spans with lightweight LLM guidance; (2) \textbf{GCLoss}, a gradient-constrained objective that concentrates token-level saliency on toxic evidence while suppressing irrelevant activations; and (3) \textbf{ARCL}, which constructs sample-specific contrastive reasoning pairs to sharpen the semantic boundary between toxic and non-toxic content. Experiments show that ToxiTrace improves classification accuracy and toxic span extraction while preserving efficient encoder-based inference and producing more coherent, human-readable explanations. We have released the model at https://huggingface.co/ArdLi/ToxiTrace.

评分理由: 论文主要研究使用蛋白质语言模型（ProtBERT）等表示方法进行帕金森病分类，属于生物信息学应用。与关键词"AI for Science” OR “Bioinformatics” OR “Cheminformatics"高度相关（10分），因为论文明确应用AI于生物医学领域。与"Large Language Models” OR “LLMs” OR “Foundation Models"有一定关联（5分），因为ProtBERT是基于Transformer的蛋白质语言模型，属于大语言模型在科学领域的应用变体。其他关键词如MoE、SLMs、Scaling Laws、训练方法、推理优化、代理系统等均未在论文中涉及或提及，因此评分为0分。

由于帕金森病的多因素性质，寻找其可靠的分子生物标志物仍然面临挑战。尽管蛋白质序列是基础且广泛可用的生物信息来源，但其自身对于复杂疾病分类的判别能力尚不明确。本研究对仅从蛋白质一级序列衍生的多种表征进行了受控且无数据泄露的评估，包括氨基酸组成、k-mer、理化描述符、混合表征以及来自蛋白质语言模型的嵌入表示。所有评估均在嵌套分层交叉验证框架下进行，以确保性能估计的无偏性。最佳性能配置（ProtBERT + MLP）取得了0.704 +/- 0.028的F1分数和0.748 +/- 0.047的ROC-AUC值，表明其判别性能仅为中等水平。经典表征如k-mer达到了可比的F1值（最高约0.667），但表现出高度不平衡的行为：召回率接近0.98，而精确度约为0.50，反映出对阳性预测的强烈偏向。在所有表征中，性能差异均保持在较窄范围内（F1分数介于0.60至0.70之间），而无监督分析显示不存在与类别标签对齐的内在结构，统计检验（弗里德曼检验，p = 0.1749）也未表明模型间存在显著差异。这些结果证明类别间存在实质性重叠，并表明仅凭一级序列信息对帕金森病分类的判别能力有限。本研究建立了一个可复现的基线，并提供了实证证据，表明需要更具信息量的生物学特征（如结构、功能或基于相互作用的描述符）来构建稳健的疾病模型。

The identification of reliable molecular biomarkers for Parkinson’s disease remains challenging due to its multifactorial nature. Although protein sequences constitute a fundamental and widely available source of biological information, their standalone discriminative capacity for complex disease classification remains unclear. In this work, we present a controlled and leakage-free evaluation of multiple representations derived exclusively from protein primary sequences, including amino acid composition, k-mers, physicochemical descriptors, hybrid representations, and embeddings from protein language models, all assessed under a nested stratified cross-validation framework to ensure unbiased performance estimation. The best-performing configuration (ProtBERT + MLP) achieves an F1-score of 0.704 +/- 0.028 and ROC-AUC of 0.748 +/- 0.047, indicating only moderate discriminative performance. Classical representations such as k-mers reach comparable F1 values (up to approximately 0.667), but exhibit highly imbalanced behavior, with recall close to 0.98 and precision around 0.50, reflecting a strong bias toward positive predictions. Across representations, performance differences remain within a narrow range (F1 between 0.60 and 0.70), while unsupervised analyses reveal no intrinsic structure aligned with class labels, and statistical testing (Friedman test, p = 0.1749) does not indicate significant differences across models. These results demonstrate substantial overlap between classes and indicate that primary sequence information alone provides limited discriminative power for Parkinson’s disease classification. This work establishes a reproducible baseline and provides empirical evidence that more informative biological features, such as structural, functional, or interaction-based descriptors, are required for robust disease modeling.

评分理由: 该论文提出了一种基于LLM的目标提示对比主题建模方法（GCTM-OT），核心创新在于将人类目标整合到主题建模中，使用LLM进行目标候选提取，并通过最优传输进行对比学习。论文明确提到"LLM-based approaches"和"LLM-based prompting”，因此与"Large Language Models"高度相关（10分）。其他关键词如MoE、SLMs、Scaling Laws、各种训练技术（预训练、微调、对齐等）、推理优化、代理系统、模型压缩、科学AI等均未在论文中涉及，因此评分为0分。

现有主题建模方法（从LDA到近期基于神经网络与大语言模型的技术）主要关注统计连贯性，往往产生冗余或偏离目标的话题，未能捕捉用户的深层意图。我们提出人本主题建模（Human-centric Topic Modeling, Human-TM），这是一种新颖的任务框架，将人类提供的目标直接整合到主题建模过程中，以生成可解释、多样化且面向目标的话题。为应对这一挑战，我们提出基于最优传输的目标提示对比主题模型（Goal-prompted Contrastive Topic Model with Optimal Transport, GCTM-OT）。该方法首先利用基于大语言模型的提示技术从文档中提取目标候选，随后通过最优传输将其融入语义感知的对比学习框架中进行主题发现。在三个公开的Reddit子论坛数据集上的实验结果表明，GCTM-OT在主题连贯性与多样性方面优于当前最先进的基线模型，同时显著提升了与人类设定目标的契合度，为人本主题发现系统的未来发展开辟了新路径。

Existing topic modeling methods, from LDA to recent neural and LLM-based approaches, which focus mainly on statistical coherence, often produce redundant or off-target topics that miss the user’s underlying intent. We introduce Human-centric Topic Modeling, \emph{Human-TM}), a novel task formulation that integrates a human-provided goal directly into the topic modeling process to produce interpretable, diverse and goal-oriented topics. To tackle this challenge, we propose the \textbf{G}oal-prompted \textbf{C}ontrastive \textbf{T}opic \textbf{M}odel with \textbf{O}ptimal \textbf{T}ransport (GCTM-OT), which first uses LLM-based prompting to extract goal candidates from documents, then incorporates these into semantic-aware contrastive learning via optimal transport for topic discovery. Experimental results on three public subreddit datasets show that GCTM-OT outperforms state-of-the-art baselines in topic coherence and diversity while significantly improving alignment with human-provided goals, paving the way for more human-centric topic discovery systems.

评分理由: 论文专注于多模态大语言模型（MLLMs）在空间推理任务中的性能提升，核心创新在于动态聚合多层几何特征以对齐任务需求。因此，仅与关键词"Large Language Models" OR “LLMs” OR “Foundation Models"高度相关（得10分），因为MLLMs是LLMs的扩展，且论文直接涉及3D基础模型。其他关键词如MoE、SLMs、Scaling Laws、Pre-training、Post-training、Alignment、RLHF、PEFT、RAG、Context Window、KV Cache、Reasoning、Agents、Quantization、Speculative Decoding、Hallucination、Interpretability、World Models、Model Merging、In-context Learning、AI for Science等均未在论文中提及或讨论，故得0分。

多模态大语言模型（MLLMs）在各种视觉任务中展现出卓越性能，但在空间推理方面仍面临挑战。近期研究尝试通过引入来自三维基础模型的几何特征来缓解此问题，但这些方法依赖于静态的单层特征提取。我们发现，此类方法会引发任务对齐偏差：几何特征自然趋向于三维预训练目标，这可能与多模态大语言模型多样化的空间需求相矛盾，使得任何单层特征本质上都无法满足要求。为解决这一问题，我们提出GeoAlign——一种动态聚合多层几何特征以重新对齐实际需求的新框架。GeoAlign构建了分层几何特征库，并利用多模态大语言模型原有的视觉标记作为内容感知查询，执行分层稀疏路由，自适应地为每个图像块提取合适的几何特征。在VSI-Bench、ScanQA和SQA3D数据集上的大量实验表明，我们仅40亿参数的紧凑模型有效实现了最先进的性能，甚至超越了现有规模更大的多模态大语言模型。

Multimodal large language models (MLLMs) have exhibited remarkable performance in various visual tasks, yet still struggle with spatial reasoning. Recent efforts mitigate this by injecting geometric features from 3D foundation models, but rely on static single-layer extractions. We identify that such an approach induces a task misalignment bias: the geometric features naturally evolve towards 3D pretraining objectives, which may contradict the heterogeneous spatial demands of MLLMs, rendering any single layer fundamentally insufficient. To resolve this, we propose GeoAlign, a novel framework that dynamically aggregates multi-layer geometric features to realign with the actual demands. GeoAlign constructs a hierarchical geometric feature bank and leverages the MLLM’s original visual tokens as content-aware queries to perform layer-wise sparse routing, adaptively fetching the suitable geometric features for each patch. Extensive experiments on VSI-Bench, ScanQA, and SQA3D demonstrate that our compact 4B model effectively achieves state-of-the-art performance, even outperforming larger existing MLLMs.

评分理由: 论文专注于计算机视觉领域的伪装目标检测（COD），提出了一种基于Segment Anything Model（SAM）的多模态提示学习框架。该研究主要涉及视觉模型（SAM）的参数高效微调（PEFT）技术，与关键词"PEFT” OR “LoRA” OR “Parameter-efficient Fine-tuning"高度相关（10分），因为论文明确提出了参数高效的适配方法。其他关键词均与论文内容无关（0分），因为论文未涉及大语言模型（LLMs）、MoE、缩放定律、对齐、推理、代理、量化等主题，也未直接涉及科学AI应用（如生物信息学）。

伪装目标检测（COD）旨在分割那些与复杂背景无缝融合的目标，当前研究日益关注利用额外的视觉模态，通过互补信息提升检测的鲁棒性。然而，现有方法大多依赖特定模态的架构或定制化的融合策略，这限制了其可扩展性与跨模态泛化能力。为此，我们提出了一种新颖的框架，为分割一切模型（Segment Anything Model，简称SAM）生成与模态无关的多模态提示，从而能够以参数高效的方式适应任意辅助模态，并显著提升COD任务的整体性能。具体而言，我们通过数据驱动的内容域与知识驱动的提示域之间的交互来建模多模态学习，将任务相关的线索提炼为统一的提示，用于SAM解码。我们进一步引入一个轻量级的掩码优化模块，通过融入细粒度的提示线索来校准粗粒度预测，从而获得更精确的伪装目标边界。在RGB-深度、RGB-热成像和RGB-偏振等多个基准数据集上的大量实验验证了我们所提出的模态无关框架的有效性与泛化能力。

Camouflaged Object Detection (COD) aims to segment objects that blend seamlessly into complex backgrounds, with growing interest in exploiting additional visual modalities to enhance robustness through complementary information. However, most existing approaches generally rely on modality-specific architectures or customized fusion strategies, which limit scalability and cross-modal generalization. To address this, we propose a novel framework that generates modality-agnostic multi-modal prompts for the Segment Anything Model (SAM), enabling parameter-efficient adaptation to arbitrary auxiliary modalities and significantly improving overall performance on COD tasks. Specifically, we model multi-modal learning through interactions between a data-driven content domain and a knowledge-driven prompt domain, distilling task-relevant cues into unified prompts for SAM decoding. We further introduce a lightweight Mask Refine Module to calibrate coarse predictions by incorporating fine-grained prompt cues, leading to more accurate camouflaged object boundaries. Extensive experiments on RGB-Depth, RGB-Thermal, and RGB-Polarization benchmarks validate the effectiveness and generalization of our modality-agnostic framework.

评分理由: 论文核心研究LLMs在调查问卷生成中的应用，并提出了一个自适应预算分配算法来优化人类验证资源的分配。因此，仅与关键词"Large Language Models” OR “LLMs” OR “Foundation Models"高度相关（10分），因为LLMs是研究的核心工具和背景。其他关键词涉及模型架构、训练方法、推理技术、特定应用领域（如生物信息学）等，论文未直接探讨这些方面，故均评0分。

大型语言模型（LLM）能够以低成本生成调查问卷的回复，但其可靠性在不同问题间差异显著，且在数据收集前无法预知。在调查中部署LLM仍需耗费高昂的人力响应进行验证与修正。应如何在问题间实时分配有限的人工标注预算？我们提出一种自适应分配算法，该算法能够在收集人工响应的同时，学习哪些问题对LLM而言最为困难。每个人工标注都扮演双重角色：既改进对该问题的估计，又揭示LLM在该问题上预测人工响应的准确程度。该算法将更多预算导向LLM可靠性最低的问题，且无需任何关于问题层面LLM准确性的先验知识。我们证明，随着预算增长，相对于最佳可能分配的分配差距趋近于零，并在合成数据以及一个包含68个问题、超过2000名受访者的真实调查数据集上验证了该方法。在真实调查数据上，传统均匀分配人工标注的做法相较于最优分配浪费了10–12%的预算；我们的算法将浪费降至2–6%，且当问题间在LLM预测质量上异质性更强时，其优势进一步扩大。该算法以更少的人工样本实现了与传统均匀抽样相同的估计质量，无需预实验，并得到经真实调查数据验证的正式性能保证。更广泛而言，本框架适用于任何需要在LLM可靠性未知的各项任务间分配稀缺人工监督的场景。

Large language models (LLMs) can generate survey responses at low cost, but their reliability varies substantially across questions and is unknown before data collection. Deploying LLMs in surveys still requires costly human responses for verification and correction. How should a limited human-labeling budget be allocated across questions in real time? We propose an adaptive allocation algorithm that learns which questions are hardest for the LLM while simultaneously collecting human responses. Each human label serves a dual role: it improves the estimate for that question and reveals how well the LLM predicts human responses on it. The algorithm directs more budget to questions where the LLM is least reliable, without requiring any prior knowledge of question-level LLM accuracy. We prove that the allocation gap relative to the best possible allocation vanishes as the budget grows, and validate the approach on both synthetic data and a real survey dataset with 68 questions and over 2000 respondents. On real survey data, the standard practice of allocating human labels uniformly across questions wastes 10–12% of the budget relative to the optimal; our algorithm reduces this waste to 2–6%, and the advantage grows as questions become more heterogeneous in LLM prediction quality. The algorithm achieves the same estimation quality as traditional uniform sampling with fewer human samples, requires no pilot study, and is backed by formal performance guarantees validated on real survey data. More broadly, the framework applies whenever scarce human oversight must be allocated across tasks where LLM reliability is unknown.

评分理由: 论文主要研究基于扩散模型的单图像生成方法（StructDiff），其核心创新在于结构保持和空间可控性，属于计算机视觉和生成模型领域。与绝大多数关键词（涉及大模型技术原理、训练方法、推理优化、对齐、代理等）完全无关。唯一的相关点是摘要中提到使用大语言模型（LLMs）作为评估单图像生成的新标准，因此给予"Large Language Models"等关键词5分（有一定关联，但非论文核心）。

本文提出StructDiff，一种基于单尺度扩散模型的单图像生成框架。单图像生成旨在通过捕捉源图像的内部统计特征（无需依赖外部数据），合成具有相似视觉内容的多样化样本。然而，现有方法往往难以保持结构布局，特别是对于包含大型刚性物体或严格空间约束的图像。此外，大多数方法缺乏空间可控性，难以引导生成内容的结构或位置。为解决这些挑战，StructDiff引入了自适应感受野模块以同时保持全局与局部分布。在此基础上，StructDiff将三维位置编码作为空间先验，从而能够灵活控制生成对象的位置、尺度及局部细节。据我们所知，这种空间控制能力是首次在单图像生成中探索基于位置编码的操控方法。此外，我们提出了一种基于大语言模型的新型单图像生成评估标准，该标准专门针对现有客观指标的局限性以及用户研究的高人力成本问题。StructDiff还展示了在下游任务中的广泛适用性，例如文本引导图像生成、图像编辑、外绘和绘画到图像合成。大量实验表明，StructDiff在结构一致性、视觉质量和空间可控性方面均优于现有方法。项目页面详见https://butter-crab.github.io/StructDiff/。

This paper introduces StructDiff, a generative framework based on a single-scale diffusion model for single-image generation. Single-image generation aims to synthesize diverse samples with similar visual content to the source image by capturing its internal statistics, without relying on external data. However, existing methods often struggle to preserve the structural layout, especially for images with large rigid objects or strict spatial constraints. Moreover, most approaches lack spatial controllability, making it difficult to guide the structure or placement of generated content. To address these challenges, StructDiff introduces an \textit{adaptive receptive field} module to maintain both global and local distributions. Building on this foundation, StructDiff incorporates 3D positional encoding (PE) as a spatial prior, allowing flexible control over positions, scale, and local details of generated objects. To our knowledge, this spatial control capability represents the first exploration of PE-based manipulation in single-image generation. Furthermore, we propose a novel evaluation criterion for single-image generation based on large language models (LLMs). This criterion specifically addresses the limitations of existing objective metrics and the high labor costs associated with user studies. StructDiff also demonstrates broad applicability across downstream tasks, such as text-guided image generation, image editing, outpainting, and paint-to-image synthesis. Extensive experiments demonstrate that StructDiff outperforms existing methods in structural consistency, visual quality, and spatial controllability. The project page is available at https://butter-crab.github.io/StructDiff/.

评分理由: 论文PAL专注于AI驱动的教育平台，通过多模态内容分析和自适应决策实现实时个性化学习。所有评分关键词均针对大模型/深度学习的技术原理、训练方法、推理优化、对齐技术、应用框架等具体技术细节，而本文未提及任何具体的大模型技术（如LLM、MoE、RLHF等）、训练方法或底层优化，也未涉及科学领域的AI应用（如生物信息学）。论文仅泛泛提及“AI-powered platform”和“AI-driven education”，未与任何评分关键词建立直接技术关联，因此所有关键词相关度均为0。

人工智能驱动的教育平台在个性化方面已取得一定进展，但多数仍局限于静态适应模式——如预设测验、统一进度或通用反馈——这限制了其响应学习者动态理解能力的发展。这一不足凸显了对兼具情境感知与实时适应能力的系统的需求。本文介绍PAL（Personal Adaptive Learner，个性化自适应学习者），一个将讲座视频转化为互动学习体验的人工智能平台。PAL持续分析多模态讲座内容，在教学进程中通过动态调整难度的问题与学习者互动，并依据其回答实时适配。在每个学习阶段结束时，PAL会生成个性化总结，在强化核心概念的同时，根据学习者的兴趣定制示例。通过融合多模态内容分析与自适应决策机制，PAL为响应式数字学习贡献了一个创新框架。我们的研究表明，人工智能如何突破静态个性化局限，转向实时个体化支持，从而应对智能教育领域的核心挑战。

AI-driven education platforms have made some progress in personalisation, yet most remain constrained to static adaptation–predefined quizzes, uniform pacing, or generic feedback–limiting their ability to respond to learners’ evolving understanding. This shortfall highlights the need for systems that are both context-aware and adaptive in real time. We introduce PAL (Personal Adaptive Learner), an AI-powered platform that transforms lecture videos into interactive learning experiences. PAL continuously analyzes multimodal lecture content and dynamically engages learners through questions of varying difficulty, adjusting to their responses as the lesson unfolds. At the end of a session, PAL generates a personalized summary that reinforces key concepts while tailoring examples to the learner’s interests. By uniting multimodal content analysis with adaptive decision-making, PAL contributes a novel framework for responsive digital learning. Our work demonstrates how AI can move beyond static personalization toward real-time, individualized support, addressing a core challenge in AI-enabled education.

评分理由: 论文核心研究Direct Preference Optimization (DPO)在视觉任务中的应用，提出rDPO框架，因此与’RLHF OR RLAIF OR Direct Preference Optimization OR DPO’高度相关（10分）。论文涉及偏好优化、指令调整和微调，与’Post-training OR Supervised Fine-tuning OR SFT’和’Instruction Tuning OR Alignment OR Value Alignment’有一定关联（各5分）。论文提到大模型作为评估者（如GPT-4），与’Large Language Models OR LLMs OR Foundation Models’有间接关联（5分）。其他关键词如MoE、量化、RAG等未在摘要中体现，评为0分。

直接偏好优化（DPO）的有效性依赖于能够反映多模态任务中关键质量差异的偏好数据。现有流程通常依赖于离策略扰动或基于结果的粗粒度信号，这些方法并不适用于细粒度的视觉推理。我们提出了rDPO，一个基于实例特定量规的偏好优化框架。针对每个图像-指令对，我们创建一个清单式量规，包含必要和附加标准，用以评估来自任何可能策略的响应。指令-量规池在线下构建，并在策略内数据构建过程中重复使用。在公开的奖励建模基准测试中，基于量规的提示方法显著改进了30B-A3B评判模型，使其性能接近GPT-5.4。在公开的下游基准测试中，基于量规的过滤将宏观平均值提升至82.69，而基于结果的过滤则将其从81.14降至75.82。在综合基准上评估可扩展性时，rDPO达到了61.01，显著优于风格受限基线（52.36），并超越了59.48的基础模型。这些结果表明，视觉偏好优化受益于将策略内数据构建与实例特定的标准级反馈相结合。

The effectiveness of Direct Preference Optimization (DPO) depends on preference data that reflect the quality differences that matter in multimodal tasks. Existing pipelines often rely on off-policy perturbations or coarse outcome-based signals, which are not well suited to fine-grained visual reasoning. We propose rDPO, a preference optimization framework based on instance-specific rubrics. For each image-instruction pair, we create a checklist-style rubric of essential and additional criteria to score responses from any possible policies. The instruction-rubric pool is built offline and reused during the construction of on-policy data. On public reward modeling benchmarks, rubric-based prompting massively improves a 30B-A3B judge and brings it close to GPT-5.4. On public downstream benchmarks, rubric-based filtering raises the macro average to 82.69, whereas outcome-based filtering drops it to 75.82 from 81.14. When evaluating scalability on a comprehensive benchmark, rDPO achieves 61.01, markedly outperforming the style-constrained baseline (52.36) and surpassing the 59.48 base model. Together, these results show that visual preference optimization benefits from combining on-policy data construction with instance-specific criterion-level feedback.

评分理由: 论文专注于医学影像（CT enterography）的视觉-语言建模，主要涉及LoRA（参数高效微调）和RAG（检索增强生成）技术，属于AI for Science（生物信息学）应用。其他关键词如大模型基础技术、推理方法、对齐训练等均未涉及。

计算机断层扫描（CT）小肠造影是评估炎症性肠病（IBD）的主要影像学手段，但何种表征选择最有利于该模态的自动化分析尚不明确。本研究首次针对腹部CT小肠造影进行视觉-语言迁移学习探索，并得出两项主要发现。首先，切片嵌入的平均池化能提供更好的疾病分类评估（59.2%的三分类准确率），而注意力池化则更有利于跨模态检索（文本到图像MRR为0.235）。该模式在所有测试的LoRA配置中均成立，表明两种聚合器强调了学习表征的不同特性。其次，单切片组织对比度比广泛的空间覆盖更为重要：将互补亨氏单位（Hounsfield Unit）窗口映射到RGB通道的多窗RGB编码策略，优于所有通过多平面采样增加空间覆盖的方法，且在此设置中添加冠状面和矢状面视图反而会降低分类性能。在报告生成任务中，未使用检索上下文的微调模型在严重程度分级上（within-1 accuracy）仅达到与患病率匹配的随机基线水平（70.4% vs. 71%随机），表明模型未能学习到超越类别分布的排序信息。检索增强生成（RAG）方法在所有配置中均提升了性能，评分较随机基线提高7-14个百分点，并将序数平均绝对误差（MAE）从0.98降低至0.80-0.89。本研究采用三教师伪标签框架，使得所有比较均无需专家标注。这些发现共同为这一尚未充分探索的影像模态建立了首批基线，并为构建面向容积医学影像的视觉-语言系统提供了实用指导。

Computed tomography (CT) enterography is a primary imaging modality for assessing inflammatory bowel disease (IBD), yet the representational choices that best support automated analysis of this modality are unknown. We present the first study of vision-language transfer learning on abdominal CT enterography and identify two main findings. First, mean pooling of slice embeddings gives better categorical disease assessment (59.2% three-class accuracy), whereas attention pooling gives better cross-modal retrieval (0.235 text-to-image MRR). This pattern holds across all LoRA configurations tested and suggests that the two aggregators emphasize different properties of the learned representation. Second, per-slice tissue contrast matters more than broader spatial coverage: multi-window RGB encoding, which maps complementary Hounsfield Unit windows to RGB channels, outperforms all strategies that increase spatial coverage through multiplanar sampling, and in this setting adding coronal and sagittal views reduces classification performance. For report generation, fine-tuning without retrieval context yields within-1 severity accuracy at the prevalence-matched chance level (70.4% vs.\ 71% random), suggesting little learned ordering beyond the class distribution. Retrieval-augmented generation (RAG) improves this across all configurations, scoring 7–14 percentage points above the chance baseline and improving ordinal MAE from 0.98 to 0.80–0.89. A three-teacher pseudolabel framework enables all comparisons without expert annotations. Together, these findings provide the first baselines for this underexplored modality and offer practical guidance for building vision-language systems for volumetric medical imaging.

评分理由: 论文核心研究On-policy Distillation (OPD)这一大语言模型后训练技术，因此与’Large Language Models’和’Post-training’高度相关（10分）。论文涉及模型对齐和知识传递，与’Instruction Tuning OR Alignment’有一定关联（5分）。其他关键词如MoE、SLMs、Scaling Laws、RLHF、RAG等均未在摘要中提及或与论文主题无关，故评0分。

同策略蒸馏已成为大语言模型后训练的核心技术，但其训练动态机制尚未得到充分理解。本文对同策略蒸馏的动态过程与机制进行了系统性研究。我们首先发现决定同策略蒸馏成败的两个关键条件：（一）学生模型与教师模型应具备兼容的思维模式；（二）即使思维模式一致且教师评分更高，教师仍需提供学生训练过程中未曾接触的真正新能力。我们通过弱到强逆向蒸馏验证了这些发现，表明从学生模型视角来看，同家族的1.5B与7B参数教师在分布上不可区分。在词元层面机制的探究中，我们发现成功的同策略蒸馏表现为：在学生访问状态的高概率词元上实现渐进式对齐，这些少量共享词元集合（仅占总数3%-1%）却集中了97%-99%的概率质量。我们进一步提出两种实用策略来挽救失败的同策略蒸馏：离策略冷启动和教师对齐提示选择。最后，我们揭示同策略蒸馏表面上的密集词元级奖励“免费午餐”实则存在代价，这引发了同策略蒸馏能否扩展到长程蒸馏场景的根本性问题。

On-policy distillation (OPD) has become a core technique in the post-training of large language models, yet its training dynamics remain poorly understood. This paper provides a systematic investigation of OPD dynamics and mechanisms. We first identify that two conditions govern whether OPD succeeds or fails: (i) the student and teacher should share compatible thinking patterns; and (ii) even with consistent thinking patterns and higher scores, the teacher must offer genuinely new capabilities beyond what the student has seen during training. We validate these findings through weak-to-strong reverse distillation, showing that same-family 1.5B and 7B teachers are distributionally indistinguishable from the student’s perspective. Probing into the token-level mechanism, we show that successful OPD is characterized by progressive alignment on high-probability tokens at student-visited states, a small shared token set that concentrates most of the probability mass (97%-99%). We further propose two practical strategies to recover failing OPD: off-policy cold start and teacher-aligned prompt selection. Finally, we show that OPD’s apparent free lunch of dense token-level reward comes at a cost, raising the question of whether OPD can scale to long-horizon distillation.

评分理由: 该论文研究的是运筹学领域的电动汽车路径规划问题（E-CVRP），提出了一种双层优化框架和b-LAHC算法，属于传统组合优化和启发式算法范畴。论文内容完全不涉及大模型、深度学习、AI技术原理或AI for Science应用，所有关键词均与大模型技术、AI方法或科学AI应用无关，因此所有关键词相关度均为0分。

本文通过一种双层优化框架处理电动容量约束车辆路径问题（E-CVRP），该框架根据搜索阶段将路径规划与充电决策分别或协同处理。通过分析两者间的交互作用，我们在上层引入了一个代理目标函数以引导搜索并加速收敛。本文提出了一种双层延迟接受爬山算法（b-LAHC），其运行包含三个阶段：贪婪下降、邻域探索以及最终解优化。b-LAHC采用固定参数运行，无需复杂自适应调整，同时保持轻量高效。在IEEE WCCI-2020基准测试上的大量实验表明，相较于八种先进算法，b-LAHC取得了更优或具有竞争力的性能。在固定评估预算下，该算法在小规模算例上获得了接近最优的解，并在大规模基准测试中创造了9/10项新的最佳已知结果，将现有记录平均提升了1.07%。此外，代理目标函数与完整成本之间观察到的强相关性（虽非绝对）证明了使用代理目标函数的合理性，同时仍需要两个层次的协同求解，从而验证了所提双层框架的有效性，并凸显了其在高效求解具有层次结构的大规模路径问题方面的潜力。

This paper tackles the Electric Capacitated Vehicle Routing Problem (E-CVRP) through a bilevel optimization framework that handles routing and charging decisions separately or jointly depending on the search stage. By analyzing their interaction, we introduce a surrogate objective at the upper level to guide the search and accelerate convergence. A bilevel Late Acceptance Hill Climbing algorithm (b-LAHC) is introduced that operates through three phases: greedy descent, neighborhood exploration, and final solution refinement. b-LAHC operates with fixed parameters, eliminating the need for complex adaptation while remaining lightweight and effective. Extensive experiments on the IEEE WCCI-2020 benchmark show that b-LAHC achieves superior or competitive performance against eight state-of-the-art algorithms. Under a fixed evaluation budget, it attains near-optimal solutions on small-scale instances and sets 9/10 new best-known results on large-scale benchmarks, improving existing records by an average of 1.07%. Moreover, the strong correlation (though not universal) observed between the surrogate objective and the complete cost justifies the use of the surrogate objective while still necessitating a joint solution of both levels, thereby validating the effectiveness of the proposed bilevel framework and highlighting its potential for efficiently solving large-scale routing problems with a hierarchical structure.

评分理由: 论文核心研究大语言模型（LLMs）的高效后训练（post-training）方法，特别是基于策略蒸馏（OPD）的改进。因此，与’Large Language Models’和’Post-training OR Supervised Fine-tuning OR SFT’高度相关（核心内容）。论文在数学推理和代码生成任务上评估，与’Chain of Thought OR CoT Reasoning OR Multi-step Reasoning’有一定关联（推理任务）。其他关键词如MoE、量化、RAG、对齐等未涉及，故评分为0。

在线策略蒸馏已成为大型语言模型一种高效的后训练范式。然而，标准在线策略蒸馏在整个训练过程中都需要一个实时的教师模型推理服务器，导致巨大的基础设施开销。在本研究中，我们探讨了在线策略蒸馏是否能够离线执行。一种自然的方法是预先在有监督微调生成的轨迹上计算一次教师模型的对数概率，并在训练过程中重复使用它们。然而在实践中，这种离线变体无法可靠地达到标准在线策略蒸馏的性能。为了理解这种差异，我们识别出了一个先前被忽视、但对任何在线策略蒸馏流程都至关重要的条件，我们称之为教师一致性。该条件要求在有监督微调和在线策略蒸馏中使用相同的教师模型。我们证明，违反教师一致性会引入一个不可约的梯度偏差，导致离线和在线策略蒸馏无论训练多久都会收敛到一个次优的固定点。基于这一洞见，我们提出了闪电在线策略蒸馏，这是一个离线的在线策略蒸馏框架，它通过预先在有监督微调轨迹上计算教师模型对数概率来强制保证教师一致性。这一设计完全消除了对实时教师服务器的需求。我们进一步证明，在教师一致性条件下，闪电在线策略蒸馏与标准在线策略蒸馏共享相同的最优点，同时具有有界的梯度差异和一种有助于防止策略漂移的隐式正则化效应。在数学推理和代码生成任务上的大量实验表明，闪电在线策略蒸馏以显著提升的效率实现了最先进的性能。从一个经过有监督微调初始化的Qwen3-8B-Base模型出发，闪电在线策略蒸馏仅用30个GPU小时就在AIME 2024上达到了69.9%的准确率，相比标准在线策略蒸馏实现了4.0倍的加速，并大幅降低了学术界进行大型语言模型后训练研究的门槛。

On-policy distillation (OPD) has emerged as an efficient post-training paradigm for large language models. However, standard OPD requires a live teacher inference server throughout training, resulting in substantial infrastructure overhead. In this work, we investigate whether on-policy distillation can be performed offline. A natural approach is to precompute teacher log-probabilities once over SFT rollouts and reuse them during training. In practice, however, this offline variant fails to reliably match the performance of standard OPD. To understand this discrepancy, we identify a previously overlooked condition that is critical for any OPD pipeline, which we term teacher consistency. This condition requires that the same teacher model be used for both supervised fine-tuning and OPD. We show that violating teacher consistency introduces an irreducible gradient bias, causing both offline and online OPD to converge to a suboptimal fixed point regardless of training duration. Building on this insight, we propose Lightning OPD, an offline on-policy distillation framework that enforces teacher consistency by precomputing teacher log-probabilities over SFT rollouts. This design eliminates the need for a live teacher server entirely. We further show that, under teacher consistency, Lightning OPD shares the same optimum as standard OPD, with bounded gradient discrepancy and an implicit regularization effect that helps prevent policy drift. Extensive experiments on mathematical reasoning and code generation demonstrate that Lightning OPD achieves state-of-the-art performance with significantly improved efficiency. Starting from an SFT-initialized Qwen3-8B-Base model, Lightning OPD reaches 69.9% on AIME 2024 in just 30 GPU hours, achieving a 4.0x speedup over standard OPD and substantially lowering the barrier to entry for academic research on LLM post-training.

评分理由: 论文主要研究LLMs在软件安全领域的应用，特别是用于修复逻辑漏洞。摘要明确提到’large language models (LLMs) in understanding and repairing code’，因此与’Large Language Models OR LLMs OR Foundation Models’高度相关（8分）。论文未涉及其他关键词的具体技术细节（如MoE、SFT、RAG等），也未涉及生物信息学等科学领域，因此其他关键词均为0分。

软件中的逻辑漏洞源于程序逻辑缺陷而非内存安全问题，可能导致严重的安全故障。尽管现有的自动化程序修复技术主要专注于修复内存破坏型漏洞，但由于其对漏洞代码及其预期行为的语义理解有限，这些技术在应对逻辑漏洞时面临困难。另一方面，大型语言模型在理解和修复代码方面取得的近期成果显示出良好前景。然而，目前尚缺乏系统分析此类技术处理逻辑漏洞的能力与局限性的框架。本文旨在系统评估传统修复方法与基于大型语言模型的修复方法在解决现实世界逻辑漏洞方面的表现。为支持评估工作，我们创建了首个逻辑漏洞数据集LogicDS，包含86个已分配通用漏洞披露（CVE）编号且具有实际安全影响的逻辑漏洞。同时开发了系统化评估框架LogicEval，用于评估逻辑漏洞补丁。评估结果表明，编译与测试失败主要受提示词敏感性、代码上下文丢失及补丁定位困难等因素驱动。

Logical vulnerabilities in software stem from flaws in program logic rather than memory safety, which can lead to critical security failures. Although existing automated program repair techniques primarily focus on repairing memory corruption vulnerabilities, they struggle with logical vulnerabilities because of their limited semantic understanding of the vulnerable code and its expected behavior. On the other hand, recent successes of large language models (LLMs) in understanding and repairing code are promising. However, no framework currently exists to analyze the capabilities and limitations of such techniques for logical vulnerabilities. This paper aims to systematically evaluate both traditional and LLM-based repair approaches for addressing real-world logical vulnerabilities. To facilitate our assessment, we created the first ever dataset, LogicDS, of 86 logical vulnerabilities with assigned CVEs reflecting tangible security impact. We also developed a systematic framework, LogicEval, to evaluate patches for logical vulnerabilities. Evaluations suggest that compilation and testing failures are primarily driven by prompt sensitivity, loss of code context, and difficulty in patch localization.

评分理由: 论文核心研究指令微调（Instruction Tuning）对大型语言模型（LLMs）鲁棒性的影响，发现简单的词汇约束会导致指令微调模型响应崩溃，而基础模型不受影响。因此，与’Large Language Models’和’Instruction Tuning’高度相关（10分），因为这是研究的核心对象和主题。与’Mechanistic Interpretability’有一定关联（5分），因为论文通过线性探针分析提示表示来探究崩溃机制，涉及模型行为的解释。其他关键词如MoE、SFT、RAG等均未在论文中涉及，故得0分。

经过指令微调的大语言模型能够生成有用且结构化的回答，但当受到微不足道的约束时，这种有用性是否稳健？我们发现，简单的词汇约束（禁止使用单个标点符号或常见词汇）会导致指令微调的大语言模型（LLMs）的回答质量崩溃，在三个开源模型系列和一个闭源模型（GPT-4o-mini）的成对评估中，其回答的全面性损失了14%至48%。在由GPT-4o-mini和GPT-4o评判的1,920组成对比较中，基线回答在77%至100%的情况下更受青睐。值得注意的是，GPT-4o-mini的全面性损失达31%（基线胜率为99%），这表明这种脆弱性也延伸到了商业部署的闭源模型，这与先前关于格式层面约束的研究发现相反。通过机制分析，我们将其归因于规划失败：采用两阶段生成（首先生成自由回答，再进行约束重写）可恢复59%至96%的回答长度，并且在生成开始前，对提示词表征进行线性探测（linear probe）即可预测回答长度（$R^2 = 0.51$–$0.93$），且$R^2$值在不同模型中与崩溃程度相关。同样的线性探测在基础模型上则产生负的$R^2$值，证实了指令微调创造了编码崩溃决策的表征结构。关键的是，基础模型在相同约束下并未出现系统性崩溃，其影响微小、随机且双向，这表明指令微调通过将任务能力与狭窄的表面形式模板耦合，从而创造了这种脆弱性。该效应在MT-Bench的所有八个任务类别中均得到复现。我们进一步发现，标准的独立LLM-as-judge评估仅检测到平均3.5%的质量下降，而成对评估则揭示了23%的下降，这暴露了当前评估受限生成方法中的一个方法论盲点。

Instruction-tuned large language models produce helpful, structured responses, but how robust is this helpfulness when trivially constrained? We show that simple lexical constraints (banning a single punctuation character or common word) cause instruction-tuned LLMs to collapse their responses, losing 14–48% of comprehensiveness in pairwise evaluation across three open-weight model families and one closed-weight model (GPT-4o-mini). The baseline response is preferred in 77–100% of 1,920 pairwise comparisons judged by GPT-4o-mini and GPT-4o. Notably, GPT-4o-mini suffers 31% comprehensiveness loss (99% baseline win rate), demonstrating that the fragility extends to commercially deployed closed-weight models, contrary to prior findings on format-level constraints. Through mechanistic analysis, we identify this as a planning failure: two-pass generation (free generation followed by constrained rewriting) recovers 59–96% of response length, and linear probes on prompt representations predict response length with $R^2 = 0.51$–$0.93$ before generation begins, with $R^2$ tracking collapse severity across models. The same probes yield negative $R^2$ on base models, confirming that instruction tuning creates the representational structure encoding the collapse decision. Crucially, base models show no systematic collapse under identical constraints, with effects that are small, noisy, and bidirectional, demonstrating that instruction tuning creates this fragility by coupling task competence to narrow surface-form templates. The effect replicates on MT-Bench across all eight task categories. We further show that standard independent LLM-as-judge evaluation detects only a 3.5% average quality drop where pairwise evaluation reveals 23%, exposing a methodological blind spot in how constrained generation is assessed.

评分理由: 论文专注于NL2SQL评估指标（ROSE）的开发，这是一个自然语言处理中的特定任务评估问题。论文内容涉及SQL语义正确性评估、对抗性验证框架和评估指标设计，但完全不涉及大模型技术原理、训练方法、推理优化、对齐技术、模型压缩、AI代理或科学AI应用等关键词领域。所有关键词均与大模型技术、训练方法、优化技术或特定应用领域相关，而本文是纯粹的评估指标研究，与这些技术主题无直接关联。

执行准确率（EX）作为评估自然语言转SQL（NL2SQL）方案有效性的广泛使用指标，正变得日益不可靠。该指标对语法变化敏感，忽略了问题可能存在多种解释的可能性，且易受错误的标准答案SQL误导。为解决这一问题，我们提出了ROSE——一种以意图为中心的评估指标，其关注点在于预测的SQL是否回答了问题，而非在参考依赖范式下与标准答案SQL的一致性。ROSE采用对抗性的证明者-反驳者级联框架：SQL证明者独立评估预测SQL相对于用户意图的语义正确性，而对抗性反驳者则利用标准答案SQL作为证据，对此判断进行挑战与优化。在我们与专家对齐的验证集ROSE-VEC上，ROSE取得了与人类专家最高的一致性，其科恩卡帕系数比次优指标高出近24%。我们还对19种NL2SQL方法进行了大规模重评估，揭示了四项有价值的发现。我们公开ROSE与ROSE-VEC，以推动更可靠的NL2SQL研究。

Execution Accuracy (EX), the widely used metric for evaluating the effectiveness of Natural Language to SQL (NL2SQL) solutions, is becoming increasingly unreliable. It is sensitive to syntactic variation, ignores that questions may admit multiple interpretations, and is easily misled by erroneous ground-truth SQL. To address this, we introduce ROSE, an intent-centered metric that focuses on whether the predicted SQL answers the question, rather than consistency with the ground-truth SQL under the reference-dependent paradigm. ROSE employs an adversarial Prover-Refuter cascade: SQL Prover assesses the semantic correctness of a predicted SQL against the user’s intent independently, while Adversarial Refuter uses the ground-truth SQL as evidence to challenge and refine this judgment. On our expert-aligned validation set ROSE-VEC, ROSE achieves the best agreement with human experts, outperforming the next-best metric by nearly 24% in Cohen’s Kappa. We also conduct a largescale re-evaluation of 19 NL2SQL methods, revealing four valuable insights. We release ROSE and ROSE-VEC to facilitate more reliable NL2SQL research.

评分理由: 论文专注于自主AI代理的安全架构，与’LLM Agents/Autonomous Agents’和’Tool Use/Function Calling’高度相关（10分），因为论文讨论的是具有执行能力的代理及其安全机制。与’Large Language Models/Foundation Models’有一定关联（8分），因为AI代理通常基于大模型构建，但论文未深入讨论LLM技术细节。其他关键词（如MoE、Scaling Laws、训练方法、推理优化、科学AI应用等）与论文的安全架构主题无关，得0分。

自主人工智能代理正迅速从实验工具转变为运营基础设施，预计到2026年底，80%的企业应用将嵌入AI副驾驶。随着代理获得执行现实世界操作的能力（读取文件、运行命令、发起网络请求、修改数据库），一个根本性的安全缺口已经显现。当前主流的代理安全方法依赖于提示层护栏：即通过自然语言指令在与其试图防范的威胁相同的抽象层级进行操作。本文认为，对于具备执行能力的代理，基于提示的安全机制在架构上是不充分的，并提出了Parallax——一种基于四大原则的安全自主AI执行范式：认知-执行分离，该原则从结构上阻止推理系统执行操作；渐进确定性对抗验证，即在推理与执行之间插入一个独立的多层验证器；信息流控制，通过代理工作流传播数据敏感度标签以检测上下文相关威胁；以及可逆执行，该机制捕获破坏前的状态，以便在验证失败时实现回滚。我们推出了OpenParallax，一个用Go语言编写的开源参考实现，并采用“假定已遭入侵”评估方法进行测试，该方法完全绕过推理系统，以测试代理在完全被入侵情况下的架构边界。在涵盖九类攻击的280个对抗性测试案例中，Parallax在默认配置下成功拦截98.9%的攻击且无误报，在最高安全配置下拦截率达到100%。当推理系统被入侵时，提示层护栏无法提供任何保护，因为它们仅存在于已被入侵的系统内部；而Parallax的架构边界则始终保持有效。

Autonomous AI agents are rapidly transitioning from experimental tools to operational infrastructure, with projections that 80% of enterprise applications will embed AI copilots by the end of 2026. As agents gain the ability to execute real-world actions (reading files, running commands, making network requests, modifying databases), a fundamental security gap has emerged. The dominant approach to agent safety relies on prompt-level guardrails: natural language instructions that operate at the same abstraction level as the threats they attempt to mitigate. This paper argues that prompt-based safety is architecturally insufficient for agents with execution capability and introduces Parallax, a paradigm for safe autonomous AI execution grounded in four principles: Cognitive-Executive Separation, which structurally prevents the reasoning system from executing actions; Adversarial Validation with Graduated Determinism, which interposes an independent, multi-tiered validator between reasoning and execution; Information Flow Control, which propagates data sensitivity labels through agent workflows to detect context-dependent threats; and Reversible Execution, which captures pre-destructive state to enable rollback when validation fails. We present OpenParallax, an open-source reference implementation in Go, and evaluate it using Assume-Compromise Evaluation, a methodology that bypasses the reasoning system entirely to test the architectural boundary under full agent compromise. Across 280 adversarial test cases in nine attack categories, Parallax blocks 98.9% of attacks with zero false positives under its default configuration, and 100% of attacks under its maximum-security configuration. When the reasoning system is compromised, prompt-level guardrails provide zero protection because they exist only within the compromised system; Parallax’s architectural boundary holds regardless.

评分理由: 该论文研究的是基于强化学习（RL）的搜索代理训练框架，使用循环一致性技术来生成奖励信号，而不依赖黄金监督（如标准答案）。论文的核心是信息检索（IR）和强化学习，不涉及大模型（LLM）或深度学习技术原理的创新，也未应用于科学领域（如生物信息学）。所有关键词均与大模型、深度学习技术或科学AI应用相关，而本文专注于传统RL在搜索任务中的应用，因此与所有关键词完全无关。

强化学习（Reinforcement Learning, RL）在优化复杂信息检索任务中的搜索智能体方面展现出巨大潜力。然而，现有方法主要依赖于黄金监督信号（如标准答案），这类信号难以规模化获取。为克服这一局限，受无监督机器翻译和图像到图像转换中循环一致性技术的启发，我们提出了一种无需黄金监督的搜索智能体训练框架——循环一致性搜索（Cycle-Consistent Search, CCS）。我们的核心假设是：与不充分或无关的搜索轨迹不同，最优搜索轨迹能够无损编码问题的意图。因此，高质量的轨迹应保留准确重构原始问题所需的信息，从而为策略优化提供奖励信号。然而，简单的循环一致性目标易受信息泄漏影响，因为重构可能依赖表面的词汇线索而非底层的搜索过程。为减弱这种效应，我们引入了信息瓶颈约束，包括排除最终响应以及对搜索查询进行命名实体识别（Named Entity Recognition, NER）掩码。这些约束迫使重构过程依赖于检索到的观测结果与结构框架，确保生成的奖励信号反映信息充分性而非语言冗余性。在问答基准测试上的实验表明，CCS达到了与有监督基线相当的性能，同时优于先前不依赖黄金监督的方法。这些结果表明，在缺乏黄金监督的场景下，CCS为训练搜索智能体提供了一种可扩展的训练范式。

Reinforcement Learning (RL) has shown strong potential for optimizing search agents in complex information retrieval tasks. However, existing approaches predominantly rely on gold supervision, such as ground-truth answers, which is difficult to scale. To address this limitation, we propose Cycle-Consistent Search (CCS), a gold-supervision-free framework for training search agents, inspired by cycle-consistency techniques from unsupervised machine translation and image-to-image translation. Our key hypothesis is that an optimal search trajectory, unlike insufficient or irrelevant ones, serves as a lossless encoding of the question’s intent. Consequently, a high-quality trajectory should preserve the information required to accurately reconstruct the original question, thereby inducing a reward signal for policy optimization. However, naive cycle-consistency objectives are vulnerable to information leakage, as reconstruction may rely on superficial lexical cues rather than the underlying search process. To reduce this effect, we apply information bottlenecks, including exclusion of the final response and named entity recognition (NER) masking of search queries. These constraints force reconstruction to rely on retrieved observations together with the structural scaffold, ensuring that the resulting reward signal reflects informational adequacy rather than linguistic redundancy. Experiments on question-answering benchmarks show that CCS achieves performance comparable to supervised baselines while outperforming prior methods that do not rely on gold supervision. These results suggest that CCS provides a scalable training paradigm for training search agents in settings where gold supervision is unavailable.

评分理由: 论文明确使用LLMs进行文本到模型的翻译任务，并测试了包括chain-of-thought reasoning和agentic approaches在内的多种策略，因此与’Large Language Models’和’Chain of Thought’高度相关（10分），且agentic approaches与’LLM Agents’高度相关（10分）。其他关键词如MoE、SLMs、Scaling Laws、训练方法、优化技术、科学AI应用等均未在摘要中提及，故评0分。

利用大型语言模型（LLM）进行文本到模型的转换与优化任务正受到日益广泛的关注。本文旨在通过引入 \textsc{Text2Model} 和 \textsc{Text2Zinc} 来推进这一研究方向。\textsc{Text2Model} 是一套基于多种不同复杂度的 LLM 策略的协同辅助工具集，并包含一个在线排行榜。\textsc{Text2Zinc} 则是一个跨领域数据集，用于捕获以自然语言描述的优化与满足性问题，同时配有一个内置人工智能助手的交互式编辑器。尽管已有研究开始探索利用 LLM 将组合问题转化为形式化模型，但我们的工作是首次尝试将满足性问题与优化问题同时整合到一个统一的架构和数据集之中。此外，与现有专注于向特定求解器模型转换的研究不同，我们的方法是求解器无关的。为此，我们利用 \textsc{MiniZinc} 的求解器与范式无关的建模能力来表述组合问题。我们进行了全面的实验，比较了多种单次调用与多次调用策略的执行与求解精度，这些策略包括：零样本提示、思维链推理、通过知识图谱的中间表示、基于语法的句法编码，以及将模型分解为顺序子任务的智能体方法。我们的协同辅助策略具有竞争力，并在某些方面改进了该领域的最新研究。我们的研究结果表明，尽管 LLM 前景广阔，但它们尚未成为组合建模的“一键式”技术。我们将 \textsc{Text2Model} 协同辅助工具集与排行榜，以及 \textsc{Text2Zinc} 数据集和交互式编辑器开源，以支持缩小这一性能差距。

There is growing interest in leveraging large language models (LLMs) for text-to-model translation and optimization tasks. This paper aims to advance this line of research by introducing \textsc{Text2Model} and \textsc{Text2Zinc}. \textsc{Text2Model} is a suite of co-pilots based on several LLM strategies with varying complexity, along with an online leaderboard. \textsc{Text2Zinc} is a cross-domain dataset for capturing optimization and satisfaction problems specified in natural language, along with an interactive editor with built-in AI assistant. While there is an emerging literature on using LLMs for translating combinatorial problems into formal models, our work is the first attempt to integrate \textit{both} satisfaction and optimization problems within a \textit{unified architecture} and \textit{dataset}. Moreover, our approach is \textit{solver-agnostic} unlike existing work that focuses on translation to a solver-specific model. To achieve this, we leverage \textsc{MiniZinc}’s solver-and-paradigm-agnostic modeling capabilities to formulate combinatorial problems. We conduct comprehensive experiments to compare execution and solution accuracy across several single- and multi-call strategies, including; zero-shot prompting, chain-of-thought reasoning, intermediate representations via knowledge-graphs, grammar-based syntax encoding, and agentic approaches that decompose the model into sequential sub-tasks. Our co-pilot strategies are competitive, and in parts improve, recent research in this domain. Our findings indicate that while LLMs are promising they are not yet a push-button technology for combinatorial modeling. We contribute \textsc{Text2Model} co-pilots and leaderboard, and \textsc{Text2Zinc} and interactive editor to open-source to support closing this performance gap.