Offline PDF Q&A Bot built with Streamlit, LangChain, FAISS & Hugging Face Transformers.
- Upload any PDF and ask natural‑language questions about its content
- Local embeddings via sentence-transformers + FAISS for fast, private search
- Instruct‑style answers from a local LLM (e.g. Flan‑T5‑Base)
- Special handlers for “structural” queries (e.g. “how many lines?”)
- Dark‑mode‑friendly UI with simple, responsive layout
-
Clone the repo
git clone https://github.com/YOUR_USERNAME/askmydocs.git cd askmydocs -
Create & activate a virtual environment
python3 -m venv venv # macOS/Linux source venv/bin/activate # Windows venv\Scripts\activate
-
Install dependencies
pip install -r requirements.txt
streamlit run app.py- Open your browser to the URL shown (usually
http://localhost:8501). - Upload a PDF, type your question, and get instant answers—all offline.
.
├── app.py
├── loader.py
├── indexer.py
├── llm.py
├── qa_chain.py
├── requirements.txt
├── README.md
└── .gitignore
Feel free to open issues or submit pull requests for new features, bug‑fixes, or UI improvements!
This project is released under the MIT License.