arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.22305cs.LG
收藏

BayesFlow: 一个用于元代理辅助工作流生成的概率推断框架

BayesFlow: A Probability Inference Framework for Meta-Agent Assisted Workflow Generation

Georgia Institute of Technology(佐治亚理工学院);Amazon Web Services AI Lab(亚马逊网络服务人工智能实验室)

Bo Yuan, Yun Zhou, Zhichao Xu, Kiran Ramnath, Aosong Feng, Balasubramaniam Srinivasan

AI总结:

BayesFlow通过贝叶斯推断和并行前瞻滚动实现工作流生成的原理性改进,提升准确性达9-65个百分点。

摘要:

自动工作流生成是自动合成复杂端到端任务中LLM调用、工具调用和后处理步骤序列的过程。大多数先前方法将此任务视为一个优化问题,理论基础有限。我们提出将工作流生成视为后验分布上的贝叶斯推断,并引入贝叶斯工作流生成(BWG),一种通过并行前瞻滚动进行重要加权和顺序循环细化器进行逐步构建工作流的采样框架。我们证明,在没有细化器的情况下,加权经验分布会收敛到目标后验。我们将BWG实例化为BayesFlow,一种无需训练的工作流构建算法。在六个基准数据集上,BayesFlow在SOTA工作流生成基线上将准确性提高高达9个百分点,在零样本提示上提高高达65个百分点,确立BWG作为基于搜索的工作流设计的原理性升级。代码将在https://github.com/BoYuanVisionary/BayesFlow上提供。

英文摘要:

Automatic workflow generation is the process of automatically synthesizing sequences of LLM calls, tool invocations, and post-processing steps for complex end-to-end tasks. Most prior methods cast this task as an optimization problem with limited theoretical grounding. We propose to cast workflow generation as Bayesian inference over a posterior distribution on workflows, and introduce \textbf{Bayesian Workflow Generation (BWG)}, a sampling framework that builds workflows step-by-step using parallel look-ahead rollouts for importance weighting and a sequential in-loop refiner for pool-wide improvements. We prove that, without the refiner, the weighted empirical distribution converges to the target posterior. We instantiate BWG as \textbf{BayesFlow}, a training-free algorithm for workflow construction. Across six benchmark datasets, BayesFlow improves accuracy by up to 9 percentage points over SOTA workflow generation baselines and by up to 65 percentage points over zero-shot prompting, establishing BWG as a principled upgrade to search-based workflow design. Code will be available on https://github.com/BoYuanVisionary/BayesFlow.

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。

↑