Hello,
I would like to contribute to a tutorial on Hyperbolic Vision Transformers by Ermolov, A. et al (2022).
The paper describes a vision transformer with output embeddings mapped to the Poincare ball and optimized with a modified pairwise cross-entropy loss. It is peer-reviewed and accepted at CVPR 2022.
Thank you!