Elamur Research
☰
home / papers / 2024 / WebVoyager: Building an End-to-End Web Agent with Large Multimodal Models

WebVoyager: Building an End-to-End Web Agent with Large Multimodal Models

Written for the Research library (claude-opus-5-5) from the paper text. Original — aclanthology.org/2024.acl-long.371.pdf

Summary

Most earlier web agents read only one kind of input, usually HTML text. They are also typically tested on simulated sites or frozen page snapshots, which says little about how they would handle real browsing. The authors present WebVoyager, an agent built on a large multimodal model (GPT-4V) that carries out user requests on live websites without human help. At each step it receives two inputs: a browser screenshot where interactive elements are outlined and numbered, and brief text describing those elements. It then reasons in the style of ReAct and issues an action such as clicking, typing or scrolling. To measure performance, the team assembled a benchmark of real tasks from 15 popular sites. They also proposed using GPT-4V as an automatic judge of whole navigation trajectories. WebVoyager succeeds on 59.1% of benchmark tasks, clearly ahead of GPT-4 (All Tools) and a text-only version of itself. The automatic judge agrees with human verdicts 85.3% of the time. The work shows that visual input helps open-web agents considerably and offers a scalable way to evaluate them.

Key points

  • New benchmark of 643 tasks across 15 widely used websites, created semi-automatically with a self-instruct procedure. The agent was also tested on 90 level-1/2 web tasks from GAIA and 50 open-web tasks from SeeAct.
  • The task success rate was 59.1%, compared with 30.8% for GPT-4 (All Tools) and 40.1% for a text-only setup that uses the WebArena accessibility tree.
  • The environment runs on Selenium against real, non-hosted sites, so the agent must cope with pop-ups, ads and constantly changing content.
  • Element labeling follows Set-of-Mark prompting but relies on GPT-4V-ACT, a rule-based JavaScript tool, instead of an object detector. To limit context length, only the three latest observations are kept, while the full history of thoughts and actions is retained.
  • Main failure types were getting stuck while navigating, weak visual grounding, hallucinated answers, and not following the prompt. Current limitations include no drag actions and no video support.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Natural language processing

Cited by

109 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 106 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2024.acl-long.371

@inproceedings{he2024webvoyager,
  title = {{WebVoyager: Building an End-to-End Web Agent with Large Multimodal Models}},
  author = {He, Hongliang and Yao, Wenlin and Ma, Kaixin and Yu, Wenhao and Dai, Yong and Zhang, Hongming and Lan, Zhenzhong and Yu, Dong},
  booktitle = {Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2024},
  doi = {10.18653/v1/2024.acl-long.371},
  url = {https://doi.org/10.18653/v1/2024.acl-long.371}
}

References

50 references, 22 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Armen Aghajanyan, Bernie Huang, Candace Ross, Vladimir Karpukhin, Hu Xu, Naman Goyal, Dmytro Okhonko, Mandar Joshi, Gargi Ghosh, Mike Lewis, et al. 2022. Cm3: A causal masked multimodal model of the internet. arXiv preprint arXiv:2201.07520. — [arXiv:2201.07520](https://arxiv.org/abs/2201.07520) find
  2. AI Anthropic. 2024. Introducing the next generation of claude. find
  3. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901. find
  4. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. 2021. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374. find our card
  5. Qi Chen, Dileepa Pitawela, Chongyang Zhao, Gengze Zhou, Hsiang-Ting Chen, and Qi Wu. 2023. Webvln: Vision-and-language navigation on websites. find
  6. Kanzhi Cheng, Qiushi Sun, Yougang Chu, Fangzhi Xu, Yantao Li, Jianbing Zhang, and Zhiyong Wu. 2024. Seeclick: Harnessing gui grounding for advanced visual gui agents. find our card
  7. Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, et al. 2023. Palm: Scaling language modeling with pathways. Journal of Machine Learning Research, 24(240):1–113. find
  8. Jacob Cohen. 1960. A coefficient of agreement for nominal scales. Educational and psychological measurement, 20(1):37–46. find
  9. Yong Dai, Duyu Tang, Liangxin Liu, Minghuan Tan, Cong Zhou, Jingquan Wang, Zhangyin Feng, Fan Zhang, Xueyu Hu, and Shuming Shi. 2022. One model, multiple modalities: A sparsely activated approach for text, sound, image, video and code. arXiv preprint arXiv:2205.06126. — [arXiv:2205.06126](https://arxiv.org/abs/2205.06126) find
  10. Xiang Deng, Yu Gu, Boyuan Zheng, Shijie Chen, Samuel Stevens, Boshi Wang, Huan Sun, and Yu Su. 2023. Mind2web: Towards a generalist agent for the web. arXiv preprint arXiv:2306.06070. find our card
  11. Qingxiu Dong, Lei Li, Damai Dai, Ce Zheng, Zhiyong Wu, Baobao Chang, Xu Sun, Jingjing Xu, and Zhifang Sui. 2022. A survey for in-context learning. arXiv preprint arXiv:2301.00234. find our card
  12. Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. 2020. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929. find
  13. Joseph L Fleiss. 1971. Measuring nominal scale agreement among many raters. Psychological bulletin, 76(5):378. find
  14. Hiroki Furuta, Ofir Nachum, Kuang-Huei Lee, Yutaka Matsuo, Shixiang Shane Gu, and Izzeddin Gur. 2023. Multimodal web navigation with instruction-finetuned foundation models. arXiv preprint arXiv:2305.11854. — [arXiv:2305.11854](https://arxiv.org/abs/2305.11854) find
  15. Yash Goyal, Tejas Khot, Douglas Summers-Stay, Dhruv Batra, and Devi Parikh. 2017. Making the v in vqa matter: Elevating the role of image understanding in visual question answering. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 6904–6913. find our card
  16. Izzeddin Gur, Hiroki Furuta, Austin Huang, Mustafa Safdari, Yutaka Matsuo, Douglas Eck, and Aleksandra Faust. 2023. A real-world webagent with planning, long context understanding, and program synthesis. arXiv preprint arXiv:2307.12856. find
  17. Jack Hessel, Jena D Hwang, Jae Sung Park, Rowan Zellers, Chandra Bhagavatula, Anna Rohrbach, Kate Saenko, and Yejin Choi. 2022. The abduction of sherlock holmes: A dataset for visual abductive reasoning. In European Conference on Computer Vision, pages 558–575. Springer. find
  18. Jing Yu Koh, Robert Lo, Lawrence Jang, Vikram Duvvur, Ming Chong Lim, Po-Yu Huang, Graham Neubig, Shuyan Zhou, Ruslan Salakhutdinov, and Daniel Fried. 2024. Visualwebarena: Evaluating multimodal agents on realistic visual web tasks. find our card
  19. Kenton Lee, Mandar Joshi, Iulia Raluca Turc, Hexiang Hu, Fangyu Liu, Julian Martin Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, and Kristina Toutanova. 2023. Pix2struct: Screenshot parsing as pretraining for visual language understanding. In International Conference on Machine Learni… find
  20. Liunian Harold Li, Mark Yatskar, Da Yin, Cho-Jui Hsieh, and Kai-Wei Chang. 2019. Visualbert: A simple and performant baseline for vision and language. arXiv preprint arXiv:1908.03557. find our card
  21. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2024. Visual instruction tuning. Advances in neural information processing systems, 36. find our card
  22. Pan Lu, Swaroop Mishra, Tanglin Xia, Liang Qiu, Kai-Wei Chang, Song-Chun Zhu, Oyvind Tafjord, Peter Clark, and Ashwin Kalyan. 2022. Learn to explain: Multimodal reasoning via thought chains for science question answering. Advances in Neural Information Processing Systems, 35:2507–2521. find
  23. Pan Lu, Baolin Peng, Hao Cheng, Michel Galley, Kai-Wei Chang, Ying Nian Wu, Song-Chun Zhu, and Jianfeng Gao. 2023. Chameleon: Plug-and-play compositional reasoning with large language models. arXiv preprint arXiv:2304.09842. find our card
  24. Kaixin Ma, Hongming Zhang, Hongwei Wang, Xiaoman Pan, and Dong Yu. 2023. Laser: Llm agent with state-space exploration for web navigation. arXiv preprint arXiv:2309.08172. — [arXiv:2309.08172](https://arxiv.org/abs/2309.08172) find
  25. Grégoire Mialon, Clémentine Fourrier, Craig Swift, Thomas Wolf, Yann LeCun, and Thomas Scialom. 2023. Gaia: a benchmark for general ai assistants. arXiv preprint arXiv:2311.12983. find our card
  26. Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, et al. 2021. Webgpt: Browser-assisted question-answering with human feedback. arXiv preprint arXiv:2112.09332. find our card
  27. OpenAI. 2023. Gpt-4 technical report. find our card
  28. OpenAI. 2024. Hello gpt-4o. find
  29. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35:27730–27744. find
  30. Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, et al. 2023. Toolllm: Facilitating large language models to master 16000+ real-world apis. arXiv preprint arXiv:2307.16789. find our card
  31. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2020. Exploring the limits of transfer learning with a unified text-to-text transformer. The Journal of Machine Learning Research, 21(1):5485–5551. find
  32. Timo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu, Maria Lomeli, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. Toolformer: Language models can teach themselves to use tools. arXiv preprint arXiv:2302.04761. find our card
  33. Peter Shaw, Mandar Joshi, James Cohan, Jonathan Berant, Panupong Pasupat, Hexiang Hu, Urvashi Khandelwal, Kenton Lee, and Kristina Toutanova. 2023. From pixels to ui actions: Learning to follow instructions via graphical user interfaces. arXiv preprint arXiv:2306.00245. — [arXiv:2306.00245](https://arxiv.org/abs/2306.00245) find
  34. Tianlin Shi, Andrej Karpathy, Linxi Fan, Jonathan Hernandez, and Percy Liang. 2017. World of bits: An open-domain platform for web-based agents. In International Conference on Machine Learning, pages 3135–3144. PMLR. find
  35. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik R Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. In Thirty-seventh Conference on Neural Information Processing Systems. find our card
  36. Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805. find our card
  37. Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. 2022. Self-instruct: Aligning language model with self generated instructions. arXiv preprint arXiv:2212.10560. find
  38. Zhengxia Zou, Keyan Chen, Zhenwei Shi, Yuhong Guo, and Jieping Ye. 2023. Object detection in 20 years: A survey. Proceedings of the IEEE. find
  39. Zirui Wang, Jiahui Yu, Adams Wei Yu, Zihang Dai, Yulia Tsvetkov, and Yuan Cao. 2021. Simvlm: Simple visual language model pretraining with weak supervision. arXiv preprint arXiv:2108.10904. find our card
  40. Lilian Weng. 2023. Llm-powered autonomous agents. lilianweng.github.io. find
  41. An Yan, Zhengyuan Yang, Wanrong Zhu, Kevin Lin, Linjie Li, Jianfeng Wang, Jianwei Yang, Yiwu Zhong, Julian McAuley, Jianfeng Gao, Zicheng Liu, and Lijuan Wang. 2023. Gpt-4v in wonderland: Large multimodal models for zero-shot smartphone gui navigation. find
  42. Jianwei Yang, Hao Zhang, Feng Li, Xueyan Zou, Chunyuan Li, and Jianfeng Gao. 2023a. Set-of-mark prompting unleashes extraordinary visual grounding in gpt-4v. arXiv preprint arXiv:2310.11441. find
  43. Zhengyuan Yang, Linjie Li, Kevin Lin, Jianfeng Wang, Chung-Ching Lin, Zicheng Liu, and Lijuan Wang. 2023b. The dawn of lmms: Preliminary explorations with gpt-4v (ision). arXiv preprint arXiv:2309.17421, 9(1). find our card
  44. Shunyu Yao, Howard Chen, John Yang, and Karthik Narasimhan. 2022a. Webshop: Towards scalable real-world web interaction with grounded language agents. find our card
  45. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2022b. React: Synergizing reasoning and acting in language models. arXiv preprint arXiv:2210.03629. find our card
  46. Da Yin, Faeze Brahman, Abhilasha Ravichander, Khyathi Chandu, Kai-Wei Chang, Yejin Choi, and Bill Yuchen Lin. 2023. Lumos: Learning agents with unified data, modular design, and open-source llms. arXiv preprint arXiv:2311.05657. find
  47. Rowan Zellers, Yonatan Bisk, Ali Farhadi, and Yejin Choi. 2019. From recognition to cognition: Visual commonsense reasoning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6720–6731. find
  48. Chi Zhang, Zhao Yang, Jiaxuan Liu, Yucheng Han, Xin Chen, Zebiao Huang, Bin Fu, and Gang Yu. 2023. Appagent: Multimodal agents as smartphone users. find
  49. Boyuan Zheng, Boyu Gou, Jihyung Kil, Huan Sun, and Yu Su. 2024. Gpt-4v (ision) is a generalist web agent, if grounded. arXiv preprint arXiv:2401.01614. find our card
  50. Shuyan Zhou, Frank F Xu, Hao Zhu, Xuhui Zhou, Robert Lo, Abishek Sridhar, Xianyi Cheng, Yonatan Bisk, Daniel Fried, Uri Alon, et al. 2023. Webarena: A realistic web environment for building autonomous agents. arXiv preprint arXiv:2307.13854. find our card