Skip to content

Latest commit

 

History

History
20 lines (17 loc) · 1.01 KB

README.md

File metadata and controls

20 lines (17 loc) · 1.01 KB

free-wang-square-lc

This is a langchain for some mysterious hackathon.

Demo

Currently, I build a sample Langchain application to see how langchain communicate with PDF file and response with help from OpenAI API.

How to start?

  1. Create .env file with your own "OPENAI_API_KEY" and "TAVILY_API_KEY".
  2. Start LangServe: python example.py
  3. Go to http://localhost:8000 to how langchain intermediate steps work. You can ask questions about DARPA internet protocol design.

Structure

  1. /sample: here comes some PDF files we want our chatbot knows.
  2. example.py: the langchain server, currently does:
    • Load PDF file
    • Convert PDF content to embedding vectors
    • Create retriever tools to help search content based on user input
    • Create agent to actually execute prompt in specific model (I use "GPT-3.5-Turbo" for test)
    • Receive user input and generate output to user
  3. client.py: currently works as a client to send question. This should be replaced with Discord Bot.