README.md

metadata

title: Jack Patel AI Assistant
emoji: 🦙
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.34.1
app_file: main.py
license: mit

🤖 Jack Patel AI Assistant

A personalized AI assistant powered by a fine-tuned TinyLlama model, trained to answer questions about Jack Patel using custom data stored in Data.json.

🌟 Features

• Personalized Responses: Trained on 150+ question-answer pairs about Jack Patel (from Data.json)
• Smart Fallback: If a match isn't found in the training data, the model generates a response using TinyLlama
• Modern UI: FastAPI-powered UI using a clean and responsive HTML form
• API Access: RESTful API endpoints for programmatic access
• Container-Ready: Deployable instantly on Hugging Face Spaces via Docker or source files

🚀 How to Use

🖥️ Web Interface

Visit the Space and type any question — get a smart response about Jack Patel.

🔌 API Usage

Make a GET request to:

/api/generate?instruction=your_question

✅ Health Check

Use:

/health

to confirm the app is running correctly.

📝 Example Questions

• "What is your name?"
• "What is your father's name?"
• "Which school did you attend?"
• "What are your technical skills?"
• "Tell me about your hobbies"

🛠️ Technical Details

• Base Model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
• Fine-Tuning: LoRA with your Data.json file containing personal Q&A pairs
• Framework: FastAPI (backend), Jinja2 (templates)
• Interface: HTML/CSS in templates/index.html
• Model Inference: PyTorch + Transformers

📁 File Structure

├── main.py                 # FastAPI application logic
├── requirements.txt        # Dependencies
├── Dockerfile             # Build config for Hugging Face Space
├── Data.json              # Personalized Q&A training data
├── templates/
│   └── index.html         # Web-based UI
├── lora_model/            # LoRA fine-tuned TinyLlama model
└── static/                # Optional static files (CSS, JS)

📊 Model Behavior

• Uses Data.json: Loads Q&A pairs at runtime for exact matching
• Model-Generated Responses: For non-exact matches, the TinyLlama model generates contextual answers
• Fallback Intelligence: Matches similar or rephrased queries using embedding similarity

🔧 Setup Instructions

🧪 Local Development

pip install -r requirements.txt
python main.py

☁️ Hugging Face Spaces

1. Upload all project files including:
   • main.py
   • Data.json
   • lora_model/ folder
   • requirements.txt
   • templates/index.html
2. Hugging Face will automatically detect and run your app.

🔒 Privacy & Security

• No data is stored or logged
• Responses are generated locally in-memory
• No external API calls
• Entirely contained in the Hugging Face Space runtime

🤝 Contributing

Want to improve this AI assistant? You can:

• Add new question-answer pairs to Data.json
• Enhance the UI
• Optimize performance and latency
• Add advanced features (like semantic search)

📄 License

This project is licensed under the MIT License.

Built with ❤️ using FastAPI, PyTorch, and Hugging Face Transformers