BRAT: Bonus oRthogonAl Token for Architecture Agnostic Textual Inversion
This work addresses the need for more flexible and efficient personalization in generative AI, though it appears incremental as it builds on existing textual inversion methods.
The paper tackles the problem of personalizing diffusion models for new subjects and styles by exploring textual inversion with a vision transformer and optimizing it using bonus tokens and orthogonality, resulting in improved adherence to source images and prompts.
Textual Inversion remains a popular method for personalizing diffusion models, in order to teach models new subjects and styles. We note that textual inversion has been underexplored using alternatives to the UNet, and experiment with textual inversion with a vision transformer. We also seek to optimize textual inversion using a strategy that does not require explicit use of the UNet and its idiosyncratic layers, so we add bonus tokens and enforce orthogonality. We find the use of the bonus token improves adherence to the source images and the use of the vision transformer improves adherence to the prompt. Code is available at https://github.com/jamesBaker361/tex_inv_plus.