立场:扩展 LLM 智能体需要基于 LLM 原语的渐近分析
Position: Scaling LLM Agents Requires Asymptotic Analysis with LLM Primitives
AI总结:
本立场论文主张,应将 LLM 前向传播作为计算成本的原子单元,通过基于 LLM 原语的渐近分析来推理和优化多智能体分解系统的效率,从而为扩展 LLM 部署释放潜力。
摘要:
将困难问题分解为子问题通常能使其更易解决且更高效。随着大型语言模型(LLM)在越来越多的能力上跨越关键可靠性阈值,人们越来越致力于将系统分解为基于 LLM 的智能体集合,每个智能体可被委派子任务。然而,这种分解(即使是自动化的)通常是直觉性的,例如基于人类如何向人类团队成员分配角色。这些角色分解距离最优有多近?本立场论文认为,需要使用 LLM 原语进行渐近分析来推理此类分解系统的效率,并且此类分析的见解将为其扩展释放机会。通过将 LLM 前向传播视为计算成本的原子单元,可以将特定 LLM 的(通常不透明的)内部工作原理与编排一组 LLM 解决难题的固有效率区分开来。换言之,若要将 LLM 的部署扩展至极限,不应将 LLM 拟人化,而应使用基于 LLM 原语的渐近分析来推理并开发将大问题分解为 LLM 智能体的更强大方法。
英文摘要:
Decomposing hard problems into subproblems often makes them easier and more efficient to solve. With large language models (LLMs) crossing critical reliability thresholds for a growing slate of capabilities, there is an increasing effort to decompose systems into sets of LLM-based agents, each of whom can be delegated sub-tasks. However, this decomposition (even when automated) is often intuitive, e.g., based on how a human might assign roles to members of a human team. How close are these role decompositions to optimal? This position paper argues that asymptotic analysis with LLM primitives is needed to reason about the efficiency of such decomposed systems, and that insights from such analysis will unlock opportunities for scaling them. By treating the LLM forward pass as the atomic unit of computational cost, one can separate out the (often opaque) inner workings of a particular LLM from the inherent efficiency of how a set of LLMs are orchestrated to solve hard problems. In other words, if we want to scale the deployment of LLMs to the limit, instead of anthropomorphizing LLMs, asymptotic analysis with LLM primitives should be used to reason about and develop more powerful decompositions of large problems into LLM agents.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。