Install requirements

Check the requirements.txt file and note that the requirements specify 2 specific packages:

pip3 install -r requirements.txt

Set environment variables

The application uses several environment variables that you'll need to set. Put them in a .env file in the root directory of the project. Here's what your .env file should look like:

#Important-this needs to come from this file or won't work
OPENAI_KEY=

Launch a local version of the Milvus DB:

sudo docker-compose up -d

Use Cases

Query using Retrieval-Augmented-Generation(RAG) pattern

Ask questions to your documents Find the relevant part Summarize with LLM

This is a barebones app to show the following:

• The RAG pattern
• Usage of a vector database, particularly a production level one(in this case Milvus)
• Simple integration of Langchain

Test dataset

This repo uses a state of the union transcript as an example.

Ask questions of the document

In order to run the app:

python src/main.py

choose 1 or 2(query or ingest)

Ingest will just load a document into the LLM and should be done first.

Query will allow asking questions of that document and should be done after.

And wait for the script to require your input.

> Enter a query:

Type exit to finish the script.