📄 arXiv 论文速递

💡 该研究将智能体经验编译为持久化技能知识，解决技能自动发现与复用难题，推动智能体能力持续进化，对构建可扩展的自主系统具有重要意义。

Agent skills package specialized knowledge and workflows into reusable resources that extend AI agent capabilities. Recent work automatically discovers such skills from agent exper…

💡 针对软件工程任务中大规模轨迹数据构建成本高的问题，提出用更少轨迹实现更优性能的方法，提升大模型解决真实软件问题的效率，具有实际工程价值。

To improve large language models' ability to resolve real-world software issues, prior work has focused on constructing large-scale agent trajectory datasets and performing supervi…

💡 现有代码审查基准多为静态，无法反映真实迭代过程。该工作提出动态基准，填补评估空白，推动代码审查模型向实用化发展，对软件质量保障至关重要。

In real-world software development, code review typically involves iterative interactions between developers and reviewers to improve software quality, making the process costly an…

💡 针对LLM智能体在真实部署中的越狱风险，提出经验驱动的红队测试智能体，通过技能进化自动发现漏洞，增强产品级系统的安全性与鲁棒性。

LLM-based agents are increasingly deployed in product-level execution harnesses, where jailbreaks can trigger harmful tool use and persistent state changes, creating greater risks …

Chemical reactions are fundamentally transformations in electron space, yet most machine learning approaches model them either through \textit{de novo} generation of product molecu…

💡 在受治理组织中，LLM智能体需兼顾角色灵活演化与执行可审计性。该架构模式分离两者，解决安全与适应性冲突，为合规部署提供关键设计范式。

Large language model (LLM) agents in governed organizations must let the persona (instructions, tone, self-presentation) evolve freely, while keeping execution (stateful, audited w…

Static scanners are increasingly used to identify executable or otherwise unsafe content in machine- learning artifacts, yet conventional evaluation metrics characterize only cases…

Currently used sepsis severity indices rely on fixed variables and weights established decades ago, which are coarsely discretized and calibrated to a cohort that no longer reflect…

Friend recommendation is inherently graph-structured: the relevance of a potential connection depends on multi-hop social context rather than user attributes alone. However, deploy…

State-of-the-art action-conditioned video models are typically restricted to a single robot embodiment, preventing them from leveraging the vast corpus of heterogeneous video data …

How do large language models (LLMs) organize moral knowledge? Models detect moral content broadly, but detection is a low bar. We ask whether they go further, distinguishing moral …

Clinical language models can achieve strong in-hospital accuracy yet fail under deployment shifts because they exploit note-specific artifacts (e.g., templates, separators, boilerp…