用大模型当裁判，让本地小模型学会自动操作界面。
DPO Learning with LLMs-Judge Signal for Computer Use Agents
- 用大模型自动评估合成操作轨迹，无需人工标注。
- 本地运行的小模型在OS-World上表现超越现有基线。
- 适合关注隐私与低资源部署的GUI自动化研究者。
计算机使用代理（CUA）是能自动与图形用户界面交互完成任务的系统。随着大视觉语言模型（VLMs）的发展，CUA取得了显著进展。然而，这些代理通常依赖云端推理，计算开销大，带来严重的隐私和可扩展性问题，尤其在个人设备上运行时。本文提出一种轻量级视觉语言模型，可在本地完整运行。为训练该紧凑代理，我们引入了LLM-as-Judge框架，自动评估并筛选合成交互轨迹，生成高质量强化学习数据，无需人工标注。在OS-World基准上的实验表明，经微调的本地模型性能优于现有基线，展示了一条通往私密、高效且泛化性强的GUI代理的可行路径。
原文摘要 · Abstract (English)
Computer use agents (CUA) are systems that automatically interact with graphical user interfaces (GUIs) to complete tasks. CUA have made significant progress with the advent of large vision-language models (VLMs). However, these agents typically rely on cloud-based inference with substantial compute demands, raising critical privacy and scalability concerns, especially when operating on personal devices. In this work, we take a step toward privacy-preserving and resource-efficient agents by developing a lightweight vision-language model that runs entirely on local machines. To train this compact agent, we introduce an LLM-as-Judge framework that automatically evaluates and filters synthetic interaction trajectories, producing high-quality data for reinforcement learning without human annotation. Experiments on the OS-World benchmark demonstrate that our fine-tuned local model outperforms existing baselines, highlighting a promising path toward private, efficient, and generalizable GUI agents.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。