What do we learn from inverting CLIP models?
This work provides insights into CLIP model behavior, highlighting biases and safety concerns for researchers and practitioners in AI.
The authors tackled the problem of understanding CLIP models by inverting them, revealing that generated images align semantically with prompts but also expose issues like gender biases and NSFW content, even for innocuous inputs.
We employ an inversion-based approach to examine CLIP models. Our examination reveals that inverting CLIP models results in the generation of images that exhibit semantic alignment with the specified target prompts. We leverage these inverted images to gain insights into various aspects of CLIP models, such as their ability to blend concepts and inclusion of gender biases. We notably observe instances of NSFW (Not Safe For Work) images during model inversion. This phenomenon occurs even for semantically innocuous prompts, like "a beautiful landscape," as well as for prompts involving the names of celebrities.