JFall 2024 - Build the best knowledge retriever for your Large Language Model.

By Jettro Coenradie and Daniel Spee

Welcome to the repository that accompanies our workshop at JFall 2024. To get most out of our workshop, please follow this readme. You need Python installed and preferably and IDE. We use PyCharm, but any other IDE will do. We will push the assignments to this repository during the workshop. You can check your setup by running the tests and through a provided script.

Setting up your environment

Python

We encourage you to use a python environment manager. Poetry makes it easy to use multiple python versions and packages. where you can switch versions per project. Read this Poetry documentation page to learn how to set up your environment. No poetry installed? Read this page to install it for your environment. Poetry installation

Setting the right version of python for the project

poetry env use 3.10

Install dependencies

poetry install

Run the tests

poetry run pytest

No poetry

Setup your venv

python3 -m venv venv
source venv/bin/activate

Install dependencies

pip install -r poetry-requirements.txt

Run the project

python app_step0_environment.py

Run the test, you should find 19 running test with status OK

python -u -m unittest discover

STOP HERE FOR PREPARATION - Loading API keys

We try to limit accessing Large Language Models and vector stores to a minimum. You do not need an LLM or vector store to learn about all the elements of the Retrieval Augmented Generation framework, except for the generation part. In the workshop we use the LLM of Open AI, which is not publicly available. We will provide you with a key to access it, if you don't have your own key.

Please use this key for the workshop only, and limit the amount of interaction, or we get blocked for exceeding our limits. The API key is obtained through a remote file, which is encrypted. Of course you can also use your own key if you have it.

Environment variables

The easiest way to load the API key is to set an environment variable for each required key. In Python we prefer the file .env in the root of the project with the following properties:

OPENAI_API_KEY=sk-...
WEAVIATE_API_KEY=...
WEAVIATE_URL=...

If you do not have your own key, you can load ours. The key is stored in a remote location. You need the .env file in the root of the project with the following line:

SECRET_KEY=...

This secret key is used to decrypt the remote file containing the API keys. We will provide the value for this key during the workshop.

Run the project

poetry run python app_step0_environment.py

Using Ollama (Optional)

There is a simple way to run a Language Model on your local machine. Depending on your machine and the chosen model, it runs fast. I am not going in to much details on how to install it, but you can find the installation instructions on the Ollama Downloads page.

At the moment we prefer the model Phi 3. You can learn more about the model on the Ollama Models page. A lot of other models are available as well. You can try them out yourself. Make sure you pull the model first before you can use it. You can also use Ollama for the embeddings. We advice to pull the model nomic-embed-text for this purpose.

ollama pull phi3
ollama pull nomic-embed-text