Chroma-RAG with OpenAI & Local Providers (Ollama)

This project integrates a Retrieval-Augmented Generation (RAG) model with Chroma, a vector database, to perform efficient and scalable search and generation tasks. You can run the system with either OpenAI's embeddings or a local embedding provider to process and retrieve relevant information from your data.

Key Features:

• Embedding Import: Import your data using embeddings into Chroma for quick and efficient search.
• Search & Query: Query the Chroma database to find the most relevant embeddings and pass them to the Large Language Model (LLM) model for generating detailed answers.
• Local vs. OpenAI Providers: Choose between local providers (e.g., Ollama) or OpenAI's API for different use cases and performance needs.
• Interactive GUI: Launch a user-friendly interface with Gradio to interact with the system easily.

Tutorial [PL]

All code and content are authored by me.

Stay tuned for the upcoming parts of my RAG tutorial series.

Available Parts

• Part 1: Import danych do bazy wektorowej

• Part 2: Przeszukiwanie bazy wektorowej

• Part 3: Tworzymy gui dla naszego chat-a

• Part 4: Dodajemy scrapowanie stron www

Install

python -m venv .venv
source .venv/bin/activate
pip3 install -r requirements.txt
cp .env_dist .env # Enter your OpenAI key

Run

Run with Local Embedding Provider and Ollama Instance

Imports data using embeddings into Chroma.

python console.py --embedder_provider local --import --verbose

Searches for similar embeddings in Chroma and passes them to the RAG.

python console.py --search_provider local --embedder_provider local --verbose --query "Show me list the projects described on blog.kamdev.pl."

Sample output

Based on the context, here are the projects listed on the blog kamdev.pl:
1. Projekt strony www (Web Project)
2. Blog fotograficzny (Photography Blog)
3. Projekt agencji masażu (Massage Agency Project)
4. Projekt wzajemniak (Unknown/ Private Project)
And also, the blog mentions a portfolio section, where you can find more projects.                                                      

Run with OpenAI Provider (Embedding & API)

Imports data using embeddings into Chroma.

python console.py --embedder_provider openai --import --verbose 

Searches for similar embeddings in Chroma and passes them to the RAG.

python console.py --search_provider openai --embedder_provider openai --verbose --query "Show me list the projects described on blog.kamdev.pl."

Sample output

The projects described on the blog kamdev.pl include: 
1. Profesionalne strony www    
2. Mój blog fotograficzny 
3. Projekt agencji masażu
4. Projekt wzajemniak
5. Agencja detektywistyczna

Run GUI with Gradio

Starts the GUI with Gradio for easier interaction.

python gui.py --embedder_provider local --verbose
python gui.py --embedder_provider openai --verbose