---
title: Mistral Rag Chat
emoji: 🏢
colorFrom: pink
colorTo: green
sdk: gradio
sdk_version: 5.34.2
app_file: app.py
pinned: false
---


# 📚 Mistral RAG Chat - Document Question Answering

**Chat with your documents using Mistral AI's powerful language models!**

Upload any text document and ask questions about its content. This app uses Retrieval-Augmented Generation (RAG) to provide accurate, context-aware answers based on your uploaded documents.

## 🚀 Features

- **📄 Document Upload**: Support for `.txt` files
- **🔍 Smart Retrieval**: Uses FAISS vector search to find relevant content
- **🤖 Mistral AI**: Powered by Mistral's large language model
- **💬 Chat Interface**: Intuitive conversation-style interaction
- **⚡ Fast Processing**: Efficient document chunking and embedding

## 🛠️ How It Works

1. **Upload** your text document (.txt format)
2. **Process** the document (creates searchable embeddings)
3. **Ask** questions about the content
4. **Get** accurate answers based on the document context

## 💡 Use Cases

- **📖 Research Papers**: Ask questions about academic papers
- **📋 Company Documents**: Query policy manuals, reports, handbooks
- **📚 Educational Content**: Study materials, textbooks, lecture notes
- **📰 News Articles**: Analyze and understand news content
- **📄 Legal Documents**: Extract key information from contracts, agreements

## 🎯 Example Queries

After uploading a document, try asking:
- "What is the main topic of this document?"
- "Summarize the key points"
- "What does the author say about [specific topic]?"
- "Are there any statistics or numbers mentioned?"
- "What conclusions does the document reach?"

## 🔧 Technical Details

- **Embedding Model**: `mistral-embed` for document vectorization
- **LLM**: `mistral-large-latest` for answer generation
- **Vector Database**: FAISS for similarity search
- **Chunk Size**: 2048 characters for optimal context
- **Retrieval**: Top-2 most relevant chunks per query

## 📝 Supported Formats

Currently supports:
- `.txt` files (UTF-8 encoded)

*More formats coming soon!*

## 🚦 Getting Started

1. Click **"Upload Text File"** and select your document
2. Click **"Process Document"** and wait for confirmation
3. Start asking questions in the chat interface
4. Get instant, context-aware answers!

## ⚠️ Important Notes

- **File Size**: Keep documents under 10MB for best performance
- **Language**: Works best with English text
- **Context**: The AI only knows what's in your uploaded document
- **Privacy**: Documents are processed temporarily and not stored permanently

## 🔐 Privacy & Security

- Your documents are processed in real-time
- No permanent storage of uploaded files
- Conversations are not logged or saved
- API calls are made securely to Mistral AI

## 🆘 Troubleshooting

**Document won't process?**
- Ensure your file is in `.txt` format
- Check that the file contains readable text
- Try a smaller file if you're having issues

**Getting irrelevant answers?**
- Make sure your question relates to the document content
- Try rephrasing your question more specifically
- Check that the document was processed successfully

**Error messages?**
- Refresh the page and try again
- Ensure your document is properly formatted
- Contact support if issues persist

## 🚀 Built With

- **[Gradio](https://gradio.app/)**: Web interface framework
- **[Mistral AI](https://mistral.ai/)**: Language model and embeddings
- **[FAISS](https://faiss.ai/)**: Vector similarity search
- **[NumPy](https://numpy.org/)**: Numerical computing
- **[Hugging Face Spaces](https://huggingface.co/spaces)**: Hosting platform

## 📊 Model Information

- **Base Model**: Mistral Large (latest version)
- **Embedding Dimension**: 1024
- **Context Window**: 32k tokens
- **Languages**: Optimized for English, supports multiple languages

## 🎨 Interface Preview

```
📚 RAG Chat Interface
Upload a text file and chat with its content!

[Upload Text File] [Process Document]

Processing Status: Document processed successfully! Split into 15 chunks.

💬 Chat:
User: What is this document about?
Assistant: Based on the uploaded document, this appears to be about...

[Your Message: Ask questions about the uploaded document...]
[Send] [Clear Chat]
```


## 📄 License

This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.


---

**Made with ❤️ using Mistral AI and Gradio**

*Try it now - upload a document and start chatting!*