Curiosity: building a Perplexity-like AI search tool using LangGraph

General Introduction

Curiosity is a program designed for exploration and experimentation, primarily using the LangGraph and FastHTML technology stacks, with the goal of building a Perplexity AI of the search product. At the heart of the project is a simple ReAct Agent, utilizing the Tavily Curiosity supports a variety of large language models (LLMs), including OpenAI's gpt-4o-mini, Groq's llama3-groq-8b-8192-tool-use-preview and the Ollama The project focuses not only on the technical implementation, but also spends a lot of time on the front-end design to ensure a high quality visual and interactive experience.

Function List

• Using LangGraph and FastHTML Technology Stacks
• Integrated Tavily Search Enhanced Text Generation
• Support for multiple LLMs, including gpt-4o-mini, llama3-groq, and llama3.1
• Provides flexible back-end switching capabilities
• The front-end is built with FastHTML and supports WebSockets streaming.

Using Help

Installation steps

1. Cloning Warehouse:

   git clone https://github.com/jank/curiosity
   

2. Make sure you have the latest Python3 interpreter.
3. Set up a virtual environment and install dependencies:

   python3 -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt
   

4. establish .env file and set the following variables:

   OPENAI_API_KEY=<your_openai_api_key>
   GROQ_API_KEY=<your_groq_api_key>
   TAVILY_API_KEY=<your_tavily_api_key>
   LANGCHAIN_TRACING_V2=true
   LANGCHAIN_ENDPOINT="https://api.smith.langchain.com"
   LANGSMITH_API_KEY=<your_langsmith_api_key>
   LANGCHAIN_PROJECT="Curiosity"
   

5. Run the project:

   python curiosity.py
   

Guidelines for use

1. Initiation of projects: Run python curiosity.py After that, the project will start and run on the local server.
2. Select LLM: Select the appropriate LLM (e.g., gpt-4o-mini, llama3-groq, or llama3.1) according to your needs.
3. Search with Tavily: Enter a query into a dialog and the ReAct Agent enhances the text generation with Tavily search.
4. Front-end Interaction: The front-end of the project is built using FastHTML and supports WebSockets streaming to ensure real-time response.

common problems

• How to switch LLM: in .env file to configure the appropriate API key and select the desired LLM when starting the project.
• WebSockets Issues: If you are experiencing problems with WebSockets closing for no apparent reason, it is recommended that you check your network connection and server configuration.