π Local RAG Chat with Documents
In this project, i wanted to recreate a local Retrieval-Augmented Generation (RAG) chatbot that can answer to questions by acquiring information from personal pdf documents.
What is Retrieval-Augmented Generation (RAG)?
Retrieval-Augmented Generation (RAG) is a technique that combines the strengths of information retrieval and natural language generation. In a RAG system, a retriever fetches relevant documents or text chunks from a database, and then a generator produces a response based on the retrieved context.
- Data Indexing
- Documents: This is the starting point where multiple documents are stored.
- Vector DB: The documents are processed and indexed into a Vector Database.
- User Query
- A user query is input into the system, which interacts with the Vector Database.
- Data Retrieval & Generation
- Top-K Chunks: The Vector Database retrieves the top-K relevant chunks based on the user query.
- LLM (Large Language Model): These chunks are then fed into a Large Language Model.
- Response: The LLM generates a response based on the relevant chunks.
ποΈ Implementation Components
For my project, i exploited the following components to build the RAG architecture:
- Chroma: A vector database used to store and retrieve document embeddings efficiently.
- Flask: Framework for rendering web page and handling user interactions.
- Ollama: Manages the local language model for generating responses.
- LangChain: A framework for integrating language models and retrieval systems.
π GitHub Repository
Visit the project repository here for accessing the codebase (if you enjoyed this content please consider leaving a star β).
π οΈ Setup and Local Deployment
- Download and install Ollama on your PC:
- Visit Ollamaβs official website to download and install Ollama. Ensure you have sufficient hardware resources to run the local language model.
- Pull a LMM of your choice:
1
ollama pull <model_name> # e.g. ollama pull llama3:8b
- Clone the repository and navigate to the project directory:
1 2
git clone https://github.com/enricollen/rag-conversational-agent.git cd rag-conversational-agent
- Create a virtual environment:
1 2
python -m venv venv source venv/bin/activate # On Windows, use `venv\Scripts\activate`
- Install the required libraries:
1
pip install -r requirements.txt
Insert you own PDFs in /data folder
- Run the application:
1
python app.py
Navigate to
http://localhost:5000/
- Perform a query
π Future Improvements
Here are some ideas for future improvements:
- Add OpenAI LLM GPT model compatibility
- Add web scraping in case none of the personal documents contain relevant info w.r.t. the query
πΉ Demo Video
Watch the demo video below to see the RAG Chatbot in action:
The demo was run on my PC with the following specifications:
- Processor: Intel(R) Core(TM) i7-14700K 3.40 GHz
- RAM: 32.0 GB
- GPU: NVIDIA GeForce RTX 3090 FE 24 GB
Screenshots πΈ
Here are some screenshots illustrating the chat: