Developed a conversational agent using a pre-trained Large Language Model (LLM) with Retrieval-Augmented Generation (RAG) to answer questions from PDF documents. Integrated LangChain, ChromaDB, Sentence Transformers, and Mistral 7B for semantic search and context-aware response generation.
A Retrieval-Augmented Generation (RAG) chatbot that answers questions based on PDF documents using Mistral, ChromaDB, and a Gradio interface.
You can use this chatbot with any PDF documents by placing them in the data/ directory and adjusting the path in the configuration.
- Processes PDF documents to create a searchable knowledge base.
- Uses Mistral (via Hugging Face) for generating clear and concise answers.
- Provides a user-friendly Gradio web interface to ask questions and view responses with sources.
- Modular code structure for easy maintenance and extension.
- Python 3.8+
- A GPU is recommended for faster model inference.
- Git installed for cloning the repository.
git clone https://github.com/ImenBenAmar/llm-rag-QA.gitpip install -r requirements.txt
-
Place your PDF files in the data/ directory (create it if it doesn't exist).
-
The default data path is data/ (configured in src/config.py).
-
If your PDFs are elsewhere, update DATA_PATH in src/config.py to point to your directory.
-
The default model is mistralai/Mixtral-8x7B-Instruct-v0.1 (Hugging Face).
-
Ensure you have internet access to download it or update MODEL_NAME in src/config.py if using a local model.
python run.py
-
A Gradio interface will launch in your browser (typically at http://localhost:7860).
-
Enter a question related to your PDF documents.
-
The chatbot will provide a concise answer and list the sources (document IDs) from the PDFs.
This chatbot is flexible and can work with any PDF documents
| Example 1 | Example 2 | Example 3 |
|---|---|---|
 |
 |
 |