🔒 Private-LLM-for-Data-Integration

A local private LLM chatbot built for data integration and data security.
It allows you to chat or upload PDFs to extract information — all without relying on the cloud or third-party services.

---

🚀 Features

• 🔐 100% local — no internet connection required
• 📄 PDF integration — ask questions directly from uploaded documents
• 🧠 Powered by open-source LLMs (downloadable from Hugging Face)
• ⚡ GPU acceleration using PyTorch + CUDA
• 🎛️ Configurable via config.yml

---

🛠️ Setup Instructions

1. Create a Virtual Environment

python -m venv venv_gpu
source venv_gpu/scripts/activate  # On Windows: venv_gpu\Scripts\activate

2. Install Dependencies

PyTorch with CUDA

Visit the official PyTorch site to choose your configuration:
👉 https://pytorch.org/get-started/locally/

Or install directly (for CUDA 12.1):

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128

Additional Packages

pip install ctransformers[cuda]
pip install chromadb[client]
pip install -r requirements.txt

---

📥 Model Setup

Download any model from TheBloke’s Hugging Face repository:
👉 https://huggingface.co/TheBloke

Place the downloaded model in the designated folder, then update the model path in config.yml.

---

▶️ Run the App

streamlit run app.py

---

Contributors:

sandipanrakshit34

Niloy-Aiml34

Mrigangana-Sarkar