Discovering the Hidden Vocabulary of DALLE-2
Machine Learning
2022-06-02 v1 Computation and Language
Cryptography and Security
Computer Vision and Pattern Recognition
Abstract
We discover that DALLE-2 seems to have a hidden vocabulary that can be used to generate images with absurd prompts. For example, it seems that \texttt{Apoploe vesrreaitais} means birds and \texttt{Contarra ccetnxniams luryca tanniounons} (sometimes) means bugs or pests. We find that these prompts are often consistent in isolation but also sometimes in combinations. We present our black-box method to discover words that seem random but have some correspondence to visual concepts. This creates important security and interpretability challenges.
Keywords
Cite
@article{arxiv.2206.00169,
title = {Discovering the Hidden Vocabulary of DALLE-2},
author = {Giannis Daras and Alexandros G. Dimakis},
journal= {arXiv preprint arXiv:2206.00169},
year = {2022}
}
Comments
6 pages, 4 figures