3 ms·
Yup. Or even more clearly with example [0]. Input request for an image of a person named "Ann Graham Lotz", it returned the exact image in the training set, s
by toss1 4y ago
Yup. Or even more clearly with example [0]. Input request for an image of a person named "Ann Graham Lotz", it returned the exact image in the training set, slightly degraded.
MIT Tech Review reports research with hundreds of similar results [1]. "The researchers, from Google, DeepMind, UC Berkeley, ETH Zürich, and Princeton, got their results by prompting Stable Diffusion and Google’s Imagen with captions for images, such as a person’s name, many times. Then they analyzed whether any of the images they generated matched original images in the model’s database. The group managed to extract over 100 replicas of images in the AI’s training set. "
[0] https://news.yahoo.com/researchers-prove-ai-art-generators-204500956.html https://news.yahoo.com/researchers-prove-ai-art-generators-2...
[1] https://www.technologyreview.com/2023/02/03/1067786/ai-models-spit-out-photos-of-real-people-and-copyrighted-images/ https://www.technologyreview.com/2023/02/03/1067786/ai-model...
- zone411 4y agoThey only found images that had been replicated 100+ times in Stable Diffusion's training set. Stable Diffusion did a poor job deduplicating.
- toss1 4y agoYup; you identify a new mode of failure! If it cannot recognize one image as identical to another, it ends up overtraining and over-generalizing on the dups.... Again, showing zero of anything resembling abstract understanding; merely a statistical correlation of blobs within each image with the text, e.g., finding the common set of pixels in each image that corresponds to "astronaut" and filtering out the rest. Obviously, this is far more useful than color-key-deletion or whole-image search. But, it isn't intelligent abstraction.