Skip to content

GenAI Lab Interactive Notebooks

Interactive Jupyter notebooks demonstrating generative AI models for computational biology.

📂 Notebook Organization

Notebooks are organized by topic:

  • diffusion/ - DDPM, DiT, latent diffusion tutorials
  • vae/ - VAE, β-VAE, CVAE examples
  • foundation_models/ - Fine-tuning and adaptation
  • datasets/ - Data loading and preprocessing

📍 Viewing Options

View rendered notebooks with proper math and plots:

Diffusion Models: - 📊 DDPM Basics - 📊 SDE Formulation - 📊 Medical Imaging Diffusion - 📊 Gene Expression Diffusion

VAE Series: - 📋 Coming soon

Advantages:

  • ✅ Reliable rendering (no GitHub timeouts)
  • ✅ Math properly displayed
  • ✅ Plots and outputs preserved
  • ✅ Mobile-friendly

Option 2: Run Locally (Interactive)

Clone and run in Jupyter:

# Clone repository
git clone https://github.com/pleiadian53/genai-lab.git
cd genai-lab/notebooks  # Use the original location

# Create environment
conda env create -f ../environment.yml
conda activate genai-lab

# Or install dependencies
pip install -r requirements.txt

# Launch Jupyter
jupyter notebook

Advantages:

  • ✅ Fully interactive
  • ✅ Modify and experiment
  • ✅ Run with your own data

Option 3: GitHub.com (Quick View)

Browse notebooks directly on GitHub: - 📓 View on GitHub

Note: GitHub's notebook renderer can be slow/unreliable for large notebooks. Use GitHub Pages for best experience.

📝 Note for Contributors

These notebooks are rendered copies from the main /notebooks/ directory in the repository.

Development Workflow:

  1. Develop in /notebooks/ (primary location)
  2. Experiment freely
  3. Iterate on code and outputs

  4. Keep messy/experimental notebooks here

  5. Polish before publishing

  6. Clear outputs or save clean outputs
  7. Add markdown explanations

  8. Test end-to-end execution

  9. Publish to /docs/notebooks/ (rendered location)

  10. Copy notebook when ready for public viewing: 
    cp notebooks/diffusion/my_tutorial.ipynb docs/notebooks/diffusion/

  11. Add to git and push

  12. Add to navigation (optional)

  13. Edit mkdocs.yml to include in nav structure
  14. Notebook is accessible by URL even without nav entry

Why Two Locations?

  • /notebooks/ — Source of truth, standard location for developers
  • /docs/notebooks/ — Rendered version for documentation site

This dual-location approach: - ✅ Keeps notebooks where developers expect them - ✅ Enables reliable rendering on GitHub Pages - ✅ Allows selective publishing (not all notebooks need to be public) - ✅ Explicit "publish" step ensures quality

Duplication is intentional and minimal - only polished notebooks are copied.

🚀 Available Notebooks

Diffusion Models ✅ Available Now!

Notebook Description Link Status
DDPM Basics Introduction to Denoising Diffusion Probabilistic Models notebooks/diffusion/01_ddpm/ ✅ Available
SDE Formulation Stochastic Differential Equations for diffusion notebooks/diffusion/02_sde_formulation/ ✅ Available
Medical Imaging Diffusion Diffusion models for medical images notebooks/diffusion/03_medical_imaging_diffusion/ ✅ Available
Gene Expression Diffusion Applying diffusion to gene expression data notebooks/diffusion/04_gene_expression_diffusion/ ✅ Available

Quick access: Navigate to "Diffusion (Technical Deep Dives)" in the top navigation tabs or sidebar!

VAE Series

Notebook Description Status
Coming soon VAE training 📋 Planned

Foundation Models

Notebook Description Status
Coming soon Fine-tuning guide 📋 Planned

💡 Tips for Learning

  • Start with documentation to understand theory
  • Run notebooks to see models in action
  • Modify and experiment to deepen understanding
  • Check examples/ for production-ready scripts

🤝 Contributing

Found a bug? Have suggestions for new notebooks?

  1. Open an issue: GitHub Issues
  2. Suggest topics: What models/techniques would help your research?
  3. Share your notebooks: Built your own? We'd love to feature them!

Ready to start? Clone the repo and explore /notebooks/! 🚀