📄 ClarityQA — Chat With Your Documents

ClarityQA is a Streamlit application that allows you to upload your PDF or text files and interact with them in natural language. It uses document chunking, vector embeddings, and a local LLM to retrieve and generate precise answers to your questions.

🚀 Demo

---

✨ Features

• 📤 Upload PDF or TXT documents
• 🧩 Automatic document chunking for efficient retrieval
• 🧠 Embedding & indexing using HuggingFace MiniLM
• 🤖 Local LLM powered by Ollama with the Mistral model
• 💬 Ask questions in natural language with contextual answers
• 🗑️ Clear session to remove vectors and reset the system

---

🛠️ Installation

1. Clone the Repository

git clone https://github.com/your-username/ClarityQA.git
cd ClarityQA

2. Create a Virtual Environment

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

3. Install Dependencies

pip install -r requirements.txt

4. Install Ollama (for the LLM)

Follow the installation instructions at Ollama Docs.

Make sure the mistral model is available:

ollama pull mistral

---

▶️ Run the App

streamlit run main.py

Then open the URL provided in the terminal (usually http://localhost:8501).

---

📂 Project Structure

├── main.py                  # Streamlit app
├── requirements.txt         # Python dependencies
├── modules/
│   ├── loader.py            # PDF/TXT loader and splitter
│   ├── vector_store.py      # Embedding & Chroma vector store logic
│   └── llm_chain.py         # LLM chain with prompt
├── temp_docs/               # Temporary uploaded files (auto-created)
├── chroma_db/               # Persistent vector database (auto-created)

• temp_docs/ and chroma_db/ are created automatically at runtime.

---

🧹 Clearing Memory

Click the “🗑️ End Session” button in the app to remove embeddings and reset the app state.
No need to manually delete folders — the app handles it.

---

🤝 Contributing

Pull requests and feature suggestions are welcome!