Retrieval Augmented Generation

This repo is a modified version from the Coursera Guided Project Introduction to RAG by Alfredo Deza from Duke University.

Dataset overview:

The well-known wine dataset.

Tech Stack overview:

1. Qdrant - in-memory vector database.
2. Sentence Transformers - embeddings creation.
3. Groq's Python API - connect to the LLM after retrieving the vectors response from Qdrant.
4. Llamafile - connect to the LLM locally (alternative to GroqAPI compatible key and endpoint)
5. Phi-2 model - using bc it is small (approx 2GB) so faster to play with. Download the model from the Llamafile repository and run it locally.

Setup your environment:

Create virtual environment:

python3 -m venv .venv
source .venv/bin/activate

Install dependencies:

.venv/bin/pip install -r requirements.txt

If this project is out of date or the req file install is acting up, here are the installs:

pip install --upgrade pip setuptools build wheel ipykernel ipywidgets jupyter pandas qdrant-client groq sentence-transformers

The groq key in a .env file will be required.

GROQ_API_KEY=<keeeeeyyyyy>

Other Courses recommended by Alfredo Deza:

Machine Learning:

• MLOps Machine Learning Operations Specialization
• Open Source Platforms for MLOps
• Python Essentials for MLOps

Data Engineering:

• Linux and Bash for Data Engineering
• Web Applications and Command-Line tools for Data Engineering
• Python and Pandas for Data Engineering
• Scripting with Python and SQL for Data Engineering