My first LLM application

My first LLM application

Date
September 11, 2023
Image credit
https://unsplash.com/photos/cvBBO4PzWPg

So, here is my first published LLM application. It's a simple & straightforward application that answers questions from users based on the content of my book "Agile Marketing: From Waterfall to Waterflow".

Technology:

  • Langchain: To load and index the dataset.
  • LLM: ChatGPT 3.5 turbo.
  • Streamlit: To deploy the application.

Access the application here: Agile Marketing QnA

Notes:

  • You might need to wait a moment. Check the top right to see if the app is "running," which indicates it's loading.
  • This is my first attempt to build something from scratch, so it might not work as expected.
  • The plan is to start with this and build on top of this, concepts that I learn through courses.

Here's a quick overview of how it functions:

  1. Import Libraries: First, ensure you've installed the required packages.
  2. Add Custom CSS: This step beautifies your app.
  3. Fetch the API Key: This is necessary for the LLM to operate.
  4. Load Your Data: Langchain offers the CSVloader library, simplifying the process of loading CSV files.
  5. Create the Index: The VectorstoreIndexCreator is utilized to establish an index for queries. This is where specific questions are directed. I won’t use a database to store vectors, since there is no need.
  6. User Input: Users can pose their questions here.
  7. Generate & Display Response: The system responds to user queries.