发表机构
Fudan University; Shanghai Jiao Tong University; Shanghai Innovation Institute（复旦大学; 上海交通大学; 上海创新研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
JevSpawn通过组合动作空间将自然语言任务规范与有限概率探索结合，利用并行生成和反馈驱动分支选择，在八个基准任务上优于七个基线和TypeSafe变体，实现更快的结构化智能体推理。
AI 中文摘要
大语言模型智能体逐词元地生成中间推理和动作，导致扩展交互变得缓慢且计算成本高昂。Jev风格模型提供有限域上的快速概率预测，但要求这些域预先指定。这一限制制约了自主任务求解，因为可用动作必须从自然语言指令中推导并通过交互进行调整。我们提出JevSpawn，一种将自然语言任务规范连接到有限概率探索的组合策略。并行动作生成与反馈驱动的分支选择、表示修正以及从保留备选方案中的恢复相结合。共享动作结构和模型前缀减少了重复生成和上下文计算，无需额外训练。在八个基准任务上，与七个智能体基线及一个TypeSafe Jev变体相比，评估表明JevSpawn是结构化智能体推理的一种有前景的方法，具有更好的任务性能和更快的导航速度。
英文摘要
LLM agents generate intermediate reasoning and actions token by token, making extended interactions slow and computationally expensive. Jev-style models offer fast probabilistic predictions over finite fields, but require those fields to be specified in advance. This requirement limits autonomous task solving, where the available actions must be derived from natural language instructions and adapted through interaction. We introduce JevSpawn, a compositional policy that connects natural language task specifications to finite probabilistic exploration. Parallel action spawning is coupled with feedback driven branch selection, representation revision, and recovery from retained alternatives. Shared action structure and model prefixes reduce repeated generation and context computation without additional training. Evaluations on eight benchmark tasks against seven agent baselines and a TypeSafe Jev variant establish JevSpawn as a promising approach to structured agentic inference, with improved task performance and faster navigation.
