Repository files navigation

📃 Paper |💻 Demo

English |简体中文

7.30._small.mp4

• 2024/11/05: 🥳 MindSearch is now deployed on Puyu! 👉Try it 👈
  • Refactored the agent module based onLagent v0.5 for better performance in concurrency.
  • Improved the UI to embody the simultaneous multi-query search.

git clone https://github.com/InternLM/MindSearchcd MindSearchpip install -r requirements.txt

Before setting up the API, you need to configure environment variables. Rename the.env.example file to.env and fill in the required values.

mv .env.example .env# Open .env and add your keys and model configurations

Setup FastAPI Server.

python -m mindsearch.app --lang en --model_format internlm_server --search_engine DuckDuckGoSearch --asy

• --lang: language of the model,en for English andcn for Chinese.

• --model_format: format of the model.

  • internlm_server for InternLM2.5-7b-chat with local server. (InternLM2.5-7b-chat has been better optimized for Chinese.)
  • gpt4 for GPT4.if you want to use other models, please modifymodels

• --search_engine: Search engine.

  • DuckDuckGoSearch for search engine for DuckDuckGo.
  • BingSearch for Bing search engine.
  • BraveSearch for Brave search web api engine.
  • GoogleSearch for Google Serper web search api engine.
  • TencentSearch for Tencent search api engine.

  Please set your Web Search engine API key as theWEB_SEARCH_API_KEY environment variable unless you are usingDuckDuckGo, orTencentSearch that requires secret id asTENCENT_SEARCH_SECRET_ID and secret key asTENCENT_SEARCH_SECRET_KEY.

• --asy: deploy asynchronous agents.

Providing following frontend interfaces,

• React

First configurate the backend URL for Vite proxy.

HOST="127.0.0.1"# modify as you needPORT=8002sed -i -r"s/target:\s*\"\"/target:\"${HOST}:${PORT}\"/" frontend/React/vite.config.ts

# Install Node.js and npm# for Ubuntusudo apt install nodejs npm# for windows# download from https://nodejs.org/zh-cn/download/prebuilt-installer# Install dependenciescd frontend/Reactnpm installnpm start

Details can be found inReact

• Gradio

python frontend/mindsearch_gradio.py

• Streamlit

streamlit run frontend/mindsearch_streamlit.py

To use a different type of web search API, modify thesearcher_type attribute in thesearcher_cfg located inmindsearch/agent/__init__.py. Currently supported web search APIs include:

• GoogleSearch
• DuckDuckGoSearch
• BraveSearch
• BingSearch
• TencentSearch

For example, to change to the Brave Search API, you would configure it as follows:

BingBrowser(searcher_type='BraveSearch',topk=2,api_key=os.environ.get('BRAVE_API_KEY','YOUR BRAVE API'))

For users who prefer to interact with the backend directly, use thebackend_example.py script. This script demonstrates how to send a query to the backend and process the response.

python backend_example.py

Make sure you have set up the environment variables and the backend is running before executing the script.

python -m mindsearch.terminal

This project is released under theApache 2.0 license.

If you find this project useful in your research, please consider cite:

@article{chen2024mindsearch,  title={MindSearch: Mimicking Human Minds Elicits Deep AI Searcher},  author={Chen, Zehui and Liu, Kuikun and Wang, Qiuchen and Liu, Jiangning and Zhang, Wenwei and Chen, Kai and Zhao, Feng},  journal={arXiv preprint arXiv:2407.20183},  year={2024}}

Explore our additional research on large language models, focusing on LLM agents.

• Lagent: A lightweight framework for building LLM-based agents
• AgentFLAN: An innovative approach for constructing and training with high-quality agent datasets (ACL 2024 Findings)
• T-Eval: A Fine-grained tool utilization evaluation benchmark (ACL 2024)