LLM_RAG_Langchain_Embedding_FAISS_Gradio: Retrieval-Augmented Generation (RAG) with Gradio UI, LangChain, Groq and OpenAI LLMs

This repository provides a modular, class-based Python project for Retrieval-Augmented Generation (RAG) using Gradio, LangChain, FAISS, HuggingFace embeddings, and Groq/OpenAI-compatible LLMs. It is designed for easy extension, best practices, and onboarding new LLM developers.

Features

• File & URL Support: Upload files (PDF, DOCX, CSV, TXT, etc.) or paste a public URL for processing.
• Model Selection: Choose from multiple LLMs Like GPT, LLama and embedding models via dropdowns in the UI. All model tokens and names are read from config.yaml.
• Persistent Vectorstore: Automatically saves and loads FAISS vectorstores for each unique file or URL, enabling fast reloading and reuse.
• Chunking & Embedding: Documents are split into overlapping chunks and embedded using the selected HuggingFace model for efficient retrieval.
• Question Answering: Ask questions about the uploaded or linked content using the selected LLM.
• Gradio UI: Simple, interactive web interface for document loading, model selection, and Q&A.
• Class-based, Pythonic Design: All core logic is encapsulated in the RAGGradioApp class for maintainability and extensibility.

Project Structure

├── LLM_RAG_Langchain_Embedding_FAISS_Gradio/
│   ├── rag_gradio_llm.py         # Main class-based RAG Gradio app (entry point)
│   ├── config.yaml               # Model/API configuration
│   ├── chunks/                   # Persistent vectorstore storage (auto-generated)
│   ├── requirements.txt          # Python dependencies
│   ├── readme.md                 # This file
│   └── ...                       # Other scripts and legacy files

Setup

1. Clone the repo:

   git clone <repo-url>
   cd LLM_RAG_Langchain_Embedding_FAISS_Gradio

2. Install dependencies:

   pip install -r requirements.txt

3. Prepare config.yaml: Example:

   llm_models:
     - name: "llama3-8b-8192"
       api_key: "YOUR_GROQ_API_KEY"
       api_base: "https://api.groq.com/openai/v1"
   embedding_models:
     - "all-MiniLM-L6-v2"

   • You can add as many LLMs and embedding models as you like.

4. Run the app:

   python rag_gradio_llm.py

Usage

• Upload a file or paste a URL in the Gradio interface.
• Select your LLM and embedding model from the dropdowns.
• Click Load Document(s) to process and index the content.
• Enter your question in the textbox and click Get Answer.

For New Coders

• The main logic is in rag_gradio_llm.py inside the RAGGradioApp class.
• All configuration is handled via config.yaml.
• The code is modular, with clear docstrings and comments for each method.
• You can extend the app by subclassing RAGGradioApp or adding new methods.
• Debug print statements are included to help trace LLM calls and QA chain creation.
• See the code comments for guidance on each step of the RAG pipeline.

Screenshot of app

Contributing

• Fork the repo and create a feature branch.
• Follow PEP8 and use class-based, modular design.
• Add docstrings and comments for clarity.
• Submit a pull request with a clear description of your changes.

License

MIT