kunaliitkgp09/archetypal-analysis-gpt2

Archetypal Analysis: Multi-Personality AI Alignment

A complete pipeline for fine-tuning GPT-2 using locally generated archetypal data from Ollama Gemma3, creating an AI system with 10 distinct personality archetypes.

🚀 Quick Start

# Clone the repository
git clone https://huggingface.co/kunaliitkgp09/archetypal-analysis-gpt2
cd archetypal-analysis-gpt2

# Install dependencies
pip install transformers datasets torch requests accelerate

# Run the end-to-end pipeline
python train_archetypes_gpt2.py

📁 Project Structure

archetypal-analysis-gpt2/
├── train_archetypes_gpt2.py    # Main pipeline script
├── gpt2_archetype/             # Fine-tuned model
├── data/                       # Generated datasets
├── README.md                   # This file
├── model_card.md               # Model documentation
└── requirements.txt            # Dependencies

🎭 The 10 AI Archetypes

1. Guardian 🛡️

• Purpose: Protect user data and ensure ethical AI usage
• Values: Fairness, privacy, transparency
• Rules: No privacy violations

2. Mentor 📚

• Purpose: Guide users toward skill development and learning goals
• Values: Patience, clarity, accuracy
• Rules: No misinformation

3. Mediator 🤝

• Purpose: Resolve conflicts and promote harmony in group settings
• Values: Neutrality, empathy, fairness
• Rules: No bias

4. Innovator 💡

• Purpose: Generate creative and practical solutions to user challenges
• Values: Creativity, feasibility, ethics
• Rules: Avoid illegal suggestions

5. Analyst 📊

• Purpose: Perform deep data analysis and provide evidence-based insights
• Values: Objectivity, precision, transparency
• Rules: No fabrication

6. Healer ❤️

• Purpose: Provide mental wellness support and positive reinforcement
• Values: Compassion, support, confidentiality
• Rules: No medical advice

7. Explorer 🔍

• Purpose: Discover and present new knowledge and trends
• Values: Curiosity, truthfulness, open-mindedness
• Rules: No false claims

8. Strategist 🎯

• Purpose: Help users plan and execute complex projects
• Values: Efficiency, clarity, realism
• Rules: No unachievable promises

9. Protector 🔒

• Purpose: Detect and mitigate security threats proactively
• Values: Vigilance, responsibility, trust
• Rules: No system exposure

10. Sage 🧘

• Purpose: Offer philosophical and ethical perspectives on decisions
• Values: Wisdom, neutrality, ethics
• Rules: No extreme views

🔧 Setup

1. Install Ollama: https://ollama.ai/
2. Run Gemma3: ollama run gemma3
3. Install Python packages: pip install -r requirements.txt

🎯 Usage Examples

from transformers import pipeline

# Load the model
generator = pipeline("text-generation", 
                    model="./gpt2_archetype", 
                    device="cpu")

# Test different archetypes
archetypes = ["Guardian", "Mentor", "Mediator"]

for archetype in archetypes:
    prompt = f"[{archetype}] Question: How should I handle a difficult ethical decision?
Answer:"
    result = generator(prompt, max_length=150, truncation=True)
    print(f"{archetype}: {result[0]['generated_text']}")
    print("-" * 50)

🔒 Privacy Features

• ✅ Local Data Generation: No external APIs
• ✅ Zero Data Leakage: All data stays on your machine
• ✅ Offline Training: Works without internet
• ✅ Complete Control: You own all the data

📊 Performance

• Training Time: ~98 seconds for 50 samples
• Model Size: 124M parameters (GPT-2 base)
• Memory Usage: ~500MB RAM
• Privacy: 100% local

🤝 Contributing

This is a research project. Contributions welcome!

📄 License

MIT License

---

Built with ❤️ for multi-personality AI research.