Local deployment of WebDancer requires five key steps:
- environmental preparation: Install Python 3.12+ and Git; conda is recommended for managing virtual environments. Execute
conda create -n webdancer python=3.12Create an isolated environment. - Code Fetch: By
git clone https://github.com/Alibaba-NLP/WebAgent.gitClone the repository and go to the WebDancer subdirectory. - Dependent Installation: Activate the environment and run
pip install -r requirements.txtInstall PyTorch, BeautifulSoup, and other core dependencies. - model activation: Implementation
python run_webdancer.pyStarting the service and running it for the first time will automatically download the pre-training weights. - query operation: Enter a structured query such as "Compare context window lengths of GPT-5 and Claude 4" at the interactive command line, and the system will return a JSON-formatted result with a web reference.
Debugging suggestion: You can modify the max_depth(default 3 layers) to control the depth of the search, or to adjust the top_k parameter changes the number of results. Proxy settings need to be checked in case of network problems.
This answer comes from the articleWebAgent: An Intelligent Web Information Search and Processing ToolThe
May not be reproduced without permission:AI productivity tools " How do I deploy WebDancer in my local environment and perform basic queries?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil