Home

Awesome

PDETraversal

ICML23 paper "Latent Traversals in Generative Models as Potential Flows"
Yue Song<sup>1,2</sup>, Andy Keller<sup>1</sup>, Nicu Sebe<sup>2</sup>, Max Welling<sup>1</sup>
<sup>1</sup>University of Amsterdam, the Netherlands <br> <sup>2</sup>University of Trento, Italy <br>

<p align="center"> <img src="workflow.jpg" width="800px"/> <br> Overview of our learned potential PDEs for latent traversal in two different experimental settings. </p>

Pre-trained GAN

Please first run checkpoint2model.py for downloading pre-trained GANs, and run anime.sh and anime_eval.sh for the training potential functions and evaluation.

Pre-trained VAE

Please first run train_vae.py to train VAEs and then run mnist.sh for training potentials.

Training VAE from scratch

Please run mnist_scratch.sh for training VAEs and potentials simultaneously.

Citation

If you think the code is helpful to your research, please consider citing our paper:

@inproceedings{song2023latent,
  title={Latent Traversals in Generative Models as Potential Flows},
  author={Song, Yue and Keller, Andy and Sebe, Nicu and Welling, Max},
  booktitle={ICML},
  year={2023},
  organization={PMLR}
}

The code is built based on WarpedGANSpace and we sincerely thank their contributions. If you have any questions or suggestions, please feel free to contact me via yue.song@unitn.it.