李代数胚上的智能体技能优化
Agentic Skill Optimization over Lie Algebroids
浏览论文内容
中文总结 AI 辅助
研究智能体技能优化问题，提出LASKO框架，将技能建模为李代数胚截面，利用李括号筛选测试替代昂贵验证，在自然语言任务因果提取中实现近15倍加速，提升技能优化速度。
中文摘要 AI 辅助
智能体系统越来越多地通过编辑技能来自我提升，如提示、评分标准、计划等。技能编辑并非向量空间中的独立坐标，其效果需在部署、验证和评估后才能观察到，不同编辑可能有相同即时可见效果但在其他方面存在差异，编辑顺序也很重要。本文介绍了一种新的技能优化框架LASKO（李代数胚技能优化）。它将带类型、锚定的Markdown技能建模为基础范畴，可用编辑策略作为带锚ρ的受控李代数胚的截面。锚将编辑策略映射到其可见的Markdown效果，核ker(ρ)表示潜在模板等结构，代数胚括号衡量非交换编辑组合。初步基准结果表明，LASKO在技能优化中实现了数量级的加速，主要是因为它在进行昂贵的验证前，先用微秒级的李括号筛选测试。在自然语言任务的因果提取中，与通过运行671B参数的DeepSeek V3.1 4位模型验证所有编辑的暴力方法相比，LASKO实现了近15倍的加速。
英文摘要
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 only after rollout, validation, and critique. Distinct edits can have the same immediate visible effect while differing in routing context, template state, guardrail scope, or future composability. The order of edits can matter as well: repairing a schema before a normalization rule need not be equivalent to applying the same edits in the reverse order. This paper introduces a new framework for skill optimization called LASKO, for Lie Algebroid SKill Optimization. LASKO models typed, anchored Markdown skills as the base category and available edit policies as sections of a controlled Lie algebroid with anchor $ρ$. The anchor maps an edit policy to its visible Markdown effect; the kernel $\ker(ρ)$ represents latent template, routing, or implementation structure; and the algebroid bracket measures noncommuting edit composition. As shown in the paper, LASKO achieves order-of-magnitude speedups in skill optimization in our preliminary benchmark results, primarily because it substitutes inexpensive Lie-bracket screening tests that run in microseconds, before investing in expensive validations that require running large language models. On a causal extraction from natural language task, LASKO achieved a speedup of almost $15 \times$ compared to a brute-force approach that validated all edits by running them through a DeepSeek V3.1 4-bit model with 671B parameters.
发表机构
- Adobe Research（Adobe研究院）
- University of Massachusetts, Amherst（马萨诸塞大学阿默斯特分校）
机构由 AI 辅助整理，请以论文原文为准。
