Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

OpenWebVoyager: Building Multimodal Web Agents via Iterative Real-World Exploration, Feedback and Optimization

About

The rapid development of large language and multimodal models has sparked significant interest in using proprietary models, such as GPT-4o, to develop autonomous agents capable of handling real-world scenarios like web navigation. Although recent open-source efforts have tried to equip agents with the ability to explore environments and continuously improve over time, they are building text-only agents in synthetic environments where the reward signals are clearly defined. Such agents struggle to generalize to realistic settings that require multimodal perception abilities and lack ground-truth signals. In this paper, we introduce an open-source framework designed to facilitate the development of multimodal web agent that can autonomously conduct real-world exploration and improve itself. We first train the base model with imitation learning to gain the basic abilities. We then let the agent explore the open web and collect feedback on its trajectories. After that, it further improves its policy by learning from well-performing trajectories judged by another general-purpose model. This exploration-feedback-optimization cycle can continue for several iterations. Experimental results show that our web agent successfully improves itself after each iteration, demonstrating strong performance across multiple test sets.

Hongliang He, Wenlin Yao, Kaixin Ma, Wenhao Yu, Hongming Zhang, Tianqing Fang, Zhenzhong Lan, Dong Yu• 2024

Related benchmarks

TaskDatasetResultRank
Online Web Task ExecutionMind2Web Cross-Task
Avg Steps per Success7
6
Online Web Task ExecutionMind2Web Cross-Web
Average Steps per Successful Task6.91
6
Online Web Task ExecutionWebVoyager
Avg Steps per Success5.07
6
Web Agent Task SuccessWebVoyager
Allrecipes Success Rate22.2
6
Showing 4 of 4 rows

Other info

Code

Follow for update