Step-by-step Jupyter Notebook tutorials for Soft-IntroVAE
- This Jupyter Notebook can be opened locally with Anaconda, or online via Google Colab.
- To run online, go to https://colab.research.google.com/ and drag-and-drop the
soft_intro_vae_code_tutorial.ipynbfile.- On Colab, note the "directory" icon on the left, figures and checkpoints are saved in this directory.
- To run the training on the image dataset, it is better to have a GPU. In Google Colab select
Runtime->Change runtime type->GPU. - You can also use NBViewer to render the notebooks: Open with Jupyter NBviewer
| File name | Purpose |
|---|---|
soft_intro_vae_2d_code_tutorial.ipynb |
Soft-IntroVAE tutorial for 2D datasets |
soft_intro_vae_image_code_tutorial.ipynb |
Soft-IntroVAE tutorial for image datasets |
soft_intro_vae_bootstrap_code_tutorial.ipynb |
Bootstrap Soft-IntroVAE tutorial for image datasets (not used in the paper) |
- General Tutorials (Jupyter Notebooks with code)