Avenir-Web通过专家融合实现精准网页元素定位，提升复杂网页任务执行能力。
Avenir-Web: Human-Experience-Imitating Multimodal Web Agents with Mixture of Grounding Experts
- 采用多专家融合的元素定位机制，解决网页交互中的误识别问题。
- 在Online-Mind2Web基准上超越开源模型，接近顶级闭源模型性能。
- 适合需要高可靠性的自动化网页操作场景，如智能客服与数据抓取。
尽管多模态大语言模型取得进展，自主网页代理在复杂动态网页界面中仍难以可靠完成长周期任务。现有代理普遍存在元素定位不准、缺乏站点特定操作知识以及长期任务追踪与记忆不稳定的问题，尤其在复杂的文档对象模型结构下更为显著。为此，我们提出Avenir-Web，一种在真实部署环境下于Online-Mind2Web基准上达到新开源最优水平的网页代理。Avenir-Web结合了多专家融合的定位机制、基于经验模仿的规划策略以融入过程先验知识，并采用任务追踪清单与自适应记忆系统，实现跨多种用户界面范式的鲁棒无缝交互。我们在Online-Mind2Web——一个严格的真实世界、以用户为中心的网页任务基准上评估该模型。结果表明，Avenir-Web显著优于先前开源代理，在性能上与顶尖闭源模型相当，确立了实时网站上可靠网页代理的新开源标杆。
原文摘要 · Abstract (English)
Despite advances in multimodal large language models, autonomous web agents still struggle to reliably execute long-horizon tasks on complex and dynamic web interfaces. Existing agents often suffer from inaccurate element grounding, the absence of site-specific procedural knowledge, and unstable long-term task tracking and memory, particularly when operating over complex Document Object Model structures. To address these limitations, we introduce Avenir-Web, a web agent that achieves a new open-source state of the art on the Online-Mind2Web benchmark in real-world deployment. Avenir-Web leverages a Mixture of Grounding Experts, Experience-Imitation Planning for incorporating procedural priors, and a task-tracking checklist combined with adaptive memory to enable robust and seamless interaction across diverse user interface paradigms. We evaluate Avenir-Web on Online-Mind2Web, a rigorous benchmark of live and user-centered web tasks. Our results demonstrate that Avenir-Web significantly surpasses prior open-source agents and attains performance parity with top-tier proprietary models, thereby establishing a new open-source state of the art for reliable web agents on live websites.
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。