A lightweight Retrieval‑Augmented Generation (RAG) chatbot that lets you:
- Upload multiple documents (TXT, Markdown, PDF¹) or crawl a web page
- Chunk & embed them with Mistral embed and store vectors in FAISS
- Ask questions that are answered strictly from your private knowledge base
- Run locally or deploy to Streamlit Community Cloud in minutes
| Tool | Version (tested) |
|---|---|
| Python | 3.9 – 3.12 |
streamlit |
≥ 1.35.0 |
mistralai SDK |
≥ 0.2.3 |
langchain‑mistralai |
≥ 0.0.10 |
faiss‑cpu |
≥ 1.8.0 |
unstructured[pdf] |
(only for PDF support, optional) |
See requirements.txt for the full, pinned list.
| Name | Purpose |
|---|---|
MISTRAL_API_KEY |
Your Mistral API key (required) |
Where to set them Local dev → in your shell (or a
.envfile) Streamlit Cloud → App → Settings → Secrets.
# 1 . clone & cd into repo
git clone https://github.com/nagmudit/RAG-Application.git
cd RAG-Application
# 2 . create virtualenv & install deps
python -m venv .venv && source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements.txt
# 3 . add your API key
export MISTRAL_API_KEY=sk‑...
# 4 . run the app
streamlit run app.pyOpen http://localhost:8501 → upload docs / paste a URL → ask questions.
App is live at https://rag-application-93zetneqmwxogjstaf83gv.streamlit.app.
- Ingest knowledge Drag‑drop files (TXT/MD/PDF) or paste a URL and click ➕ Add to KB.
- Chat Type a question in the bottom chat box; answers cite snippets from your sources.
- Manage 🗑 Clear KB deletes all stored vectors; 🧹 Clear Chat wipes the conversation.
Rate limits The app retries automatically (exponential back‑off) on 429s, but if you exceed your Mistral quota, wait or upgrade.
├─ app.py # Streamlit front‑end & back‑end logic
├─ requirements.txt # Python deps
├─ faiss_db/ # (generated) FAISS index + metadata
└─ assets/
└─ screenshot.png # optional demo image
MIT © 2025