README.md

metadata

title: Matrix AI Chat with RAG
emoji: 🕶️
colorFrom: gray
colorTo: green
sdk: streamlit
sdk_version: 1.28.0
app_file: app.py
pinned: false
license: mit

🕶️ Matrix AI Chat with RAG

Enter the Matrix - Advanced AI chat with Retrieval-Augmented Generation (RAG) and multiple AI models.

✨ Features

• Multi-Model Support: DeepSeek-R1, Llama-3.2-3B, Qwen2.5-VL-7B
• RAG System: Upload documents for context-aware responses
• Matrix UI: Cyberpunk green-on-black theme with animations
• Smart Memory: Persistent chat history and similarity search
• Session Management: Save/load conversations with unique IDs

🚀 Quick Start

1. Set HF Token: Add your Hugging Face token in Space settings
2. Select Model: Choose from available AI models in sidebar
3. Chat: Start conversing with Matrix-style streaming
4. Upload Data: Add documents via "KNOWLEDGE MATRIX" section

🔧 Configuration

Required Environment Variables

• HF_TOKEN: Your Hugging Face API token with inference access

Supported Models

• DeepSeek-R1: Advanced reasoning (Together AI)
• Llama-3.2-3B: Efficient local model
• Qwen2.5-VL-7B: Vision-language model (Hyperbolic)

📋 Usage

Basic Chat

1. Select AI model from sidebar
2. Enter message in chat input
3. Watch Matrix-style streaming response

RAG Documents

1. Open "UPLOAD DATA" in sidebar
2. Add title and content
3. Click "INJECT DATA"
4. AI will reference your documents automatically

Session Controls

• NEW JACK IN: Start fresh session
• PURGE ALL: Clear all data (double-click to confirm)

🛠️ Local Development

# Clone and install
git clone <repo-url>
pip install streamlit numpy sentence-transformers scikit-learn transformers torch huggingface-hub openai

# Set token and run
export HF_TOKEN="your_token_here"
streamlit run app.py

🚨 Troubleshooting

No HF Token Found

• Add HF_TOKEN to Space settings → Repository secrets

Model Connection Failed

• Check token validity and quota
• Verify internet connection
• Try different model

Data Not Saving

• Ensure sufficient storage space
• Check write permissions

📁 Data Storage

• data/sessions/: Chat history and metadata
• data/documents/: RAG documents and embeddings
• data/embeddings/: Vector embeddings for similarity search

All data persists between sessions. Use "PURGE ALL" to reset.

---

Welcome to the Matrix. The choice is yours. 🕶️