研究人员引入了LASKO，一个新颖的Agent技能优化框架，通过将技能建模为受控李代数内的结构化产物。该方法通过使用廉价的李括号筛选测试来过滤无效的编辑，然后在与大型语言模型进行昂贵的验证之前，从而实现更快的技能优化。初步结果显示，与使用DeepSeek V3.1模型的暴力方法相比，LASKO在因果提取任务上实现了显著的加速，包括15倍的提升。 AI
影响 这项研究可能通过降低技能优化的计算成本，从而实现更高效的Agentic AI系统的开发和部署。
排序理由 该集群包含一篇详细介绍新框架和初步基准测试结果的学术论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2607.11493v1 Announce Type: cross Abstract: Agentic systems increasingly improve themselves by editing skills: prompts, rubrics, plans, tool contracts, examples, validators, and traces. Skill edits are not independent coordinates in a vector space: they are local repairs to…
Agentic systems increasingly improve themselves by editing skills: prompts, rubrics, plans, tool contracts, examples, validators, and traces. Skill edits are not independent coordinates in a vector space: they are local repairs to structured artifacts whose effects are observed o…
Agentic systems increasingly improve themselves by editing skills: prompts, rubrics, plans, tool contracts, examples, validators, and traces. Skill edits are not independent coordinates in a vector space: they are local repairs to structured artifacts whose effects are observed o…