this is a demo of our research paper Does CLIP Know My Face?.
As multi-modal models pre-trained on web-crawled data are used in many downstream tasks (i.e. Stable Diffusion and other Text-To-Image Models), privacy is an important aspect that is currently not considered. With our paper, we propose Identity Inference Attacks (IDIA) such that users can check whether their data was used to train a certain model. We believe that this is an essential step towards more privacy and transparency for training and releasing such models.
Therefore, to run the demo faster and more efficiently, we would like to apply for a GPU grant.
I am happy to answer any further questions regarding the research!
Thanks and best whishes,
Hi @D0miH, we have assigned a gpu to this space. Note that GPU Grants are provided temporarily and might be removed after some time if the usage is very low.
To learn more about GPUs in Spaces, please check out https://huggingface.co/docs/hub/spaces-gpus
Thank you very much!