arXiv cs.AI· Boris Radovi\v{c}, Marco Canini, Veljko Pejovi\'c·· 3 天前AI 评分34
小模型该坐驾驶座吗？云编写智能体程序中本地模型作为解释器的研究
Should SLMs be in the Driver's Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs
AI 导读
研究提出一种中间架构：云端模型编写的程序通过普通函数调用，将关于用户数据的有界问题交给本地小模型回答，从而在数据不出设备的前提下完成运行时内容解释。在 AppWorld 上，该方案在全部六种骨干模型与测试分区组合中均提升纯代码智能体的成功率，最高提升 14.9pp，缩小了与云端 ReAct 循环的差距。
正文 · 原文
Abstract:The most accurate agents executing user-provided tasks are driven by large, cloud-hosted models, which solve them by reading the private data of the user. Agents powered by models small enough to run on edge devices keep that data local, but yield far worse results. A third option is to have cloud models write a program that devices execute: the data stays local, but as the program is written before any data is seen, it fails whenever a step depends on interpreting content that appears only at run time. We explore an intermediate architecture, in which the cloud-written program defers bounded questions over user data to a small local model through an ordinary function call. On AppWorld this improves the success rate of a code-only agent in all six combinations of backbone model and test partition we evaluate, by up to 14.9pp, recovering a substantial part of the gap to a cloud-hosted ReAct loop while keeping user data on the device.
| Comments: | Accepted at SLM-Agents: 1st Workshop on SLMs for Agentic Systems, NeurIPS 2026 |
| Subjects: | Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC) |
| Cite as: | arXiv:2610.14209 [cs.AI] |
| (or arXiv:2610.14209v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.14209 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Boris Radovič [view email]
[v1]
Fri, 9 Oct 2026 18:20:39 UTC (127 KB)
来源：arXiv cs.AI · arxiv.org