arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2512.03571cs.AIcs.LGcs.PL
收藏

EnCompass:通过程序执行路径搜索增强智能体编程

EnCompass: Enhancing Agent Programming with Search Over Program Execution Paths

Asari AI;MIT CSAIL(麻省理工学院计算机科学与人工智能实验室);Caltech CMS(加州理工学院 CMS)

Zhening Li, Armando Solar-Lezama, Yisong Yue, Stephan Zheng

AI总结:

EnCompass通过解耦智能体工作流逻辑与推理策略,提供一种基于Python的框架,允许快速提升智能体可靠性并灵活切换推理策略。

摘要:

我们介绍了一种新的智能体编程方法,即基于大语言模型(LLM)的智能体开发。当前的智能体编程方法往往将智能体设计的两个方面纠缠在一起:核心工作流逻辑和推理时的策略(例如树搜索)。我们引入了“概率天使非确定性”(PAN),一种编程模型,将这两个方面解耦,允许程序员描述智能体工作流,并通过简单地更改几个输入独立地尝试不同的推理策略。我们以Python实现PAN作为EnCompass框架,该框架使用Python装饰器将智能体工作流程序编译为搜索空间。我们展示了三个案例研究,展示了该框架如何让程序员快速提高智能体的可靠性,并轻松切换不同的推理策略,所有这些都只需少量额外的编码。

英文摘要:

We introduce a new approach to agent programming, the development of LLM-based agents. Current approaches to agent programming often entangle two aspects of agent design: the core workflow logic and the inference-time strategy (e.g., tree search). We introduce "probabilistic angelic nondeterminism" ("PAN"), a programming model that disentangles these two concerns, allowing the programmer to describe the agent workflow and independently experiment with different inference-time strategies by simply changing a few inputs. We provide an implementation of PAN in Python as the EnCompass framework, which uses a Python decorator to compile agent workflow programs into a search space. We present three case studies that demonstrate how the framework lets the programmer quickly improve the reliability of an agent and easily switch between different inference-time strategies, all with little additional coding.

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。

↑