0

first of sorry if my question would result stupid or have no so much sense, i don't have so much experience with langchain and in general with LLM.

So, i have the following scenario

i need to use as an Embedding a book ( 300 pages ) to instruct OpenAI how to use and how to get information from a list of requirements.

I need to pass as well a very large list o requirements as an input source

and i need to ask to the LLM to analyze the list of the requirements, get the metodologies from the document for the analysis and return a list of result derived from that analysis

For example:

List of requirements:

  1. the car must have two doors
  2. every car door have at least one window

The book content to define if a car is a car for family or a sporting car, is important too check the doors number.

The question: Give a list of requirements please could you provide me which kind of car and maybe a list of already existing cars in the market.

The example is quite stupid, i know, but is an example to understand that i have different sources and must be treated in different ways.

So, my question:

Is that possible to do in langchain or with any kind of library? if yes could you help me providing an example?

I have already do some test using ChromaDB but the problem is that i have only one database for all the information and the results are very very bad

thanks in advance

Andrea Mucci
  • 747
  • 2
  • 9
  • 25
  • Consider doing more basic LangChain tutorials to understand it and discover whether or not it can help with your project. There are many Youtube videos about LangChain, OpenAI, and HuggingFace. You might need to take ideas from several different videos to achieve your goal. Like you, I'm trying to find out if/how I can do data processing & query tasks with AI. Currently, I'm learning from video "LangChain Cookbook" https://www.youtube.com/watch?v=2xxziIWmaSA. Good luck with your project. – devdanke Aug 05 '23 at 16:07

1 Answers1

0

Look at the LangChain: Chat with Your Data free short course here.

Helped me a lot!