Skip to content

Latest commit

 

History

History
17 lines (10 loc) · 999 Bytes

README.md

File metadata and controls

17 lines (10 loc) · 999 Bytes

Super quick PDF-AI parsing server in TS

TLDR

Minimalistic (for example only - not production use) demonstration of how to use Langchain/OpenAI API for a PDF server in JS & TS

  • embed.ts contains the logic for parsing, embedding, and saving a PDF document as a local vector store using FAISS
  • query.ts handles incoming query requests and returns a response using GPT3.5 and the previously saved DB info
  • /embeddings/ directory is the local directory for where our vector embeddings are saved.
  • CustomChain.ts is a custom Langchain Chain that is resposible for document retrieval, context addition, as well as the actual prompt for querying the Assistant once we retrieve the relevant information from the db.

Server

This repo uses a minimal fastify server to just demonstrate an API route for parsing, saving, and querying PDF documents with Langchain + GPT3.5

Compile the typescript and run the server.js file or run it with ts-node. It will run on localhost 3000.