Skip to content

Latest commit

 

History

History
20 lines (13 loc) · 919 Bytes

README.md

File metadata and controls

20 lines (13 loc) · 919 Bytes
Raw

Cross GAN Auditing: Unsupervised Identification of Attribute Level Similarities and Differences between Pretrained Generative Models

This repo (based off the LatentCLR repo) contains the raw code used for all experiments. It's current state is difficult to use, so please email me if you plan on running experiments and I will help you set it up. Needs hydra-core==1.0.6 and StyleGAN2-ada legacy.py conversion code.

Citation

Please cite our paper!

@inproceedings{olson2023cross,
  title={Cross-GAN Auditing: Unsupervised Identification of Attribute Level Similarities and Differences between Pretrained Generative Models},
  author={Olson, Matthew L and Liu, Shusen and Anirudh, Rushil and Thiagarajan, Jayaraman J and Bremer, Peer-Timo and Wong, Weng-Keen},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7981--7990},
  year={2023}
}