We are presenting FAE at #CVPR2026 morning 7:30-9:00 am today as well!
#134
We found a better latent space for generative models with single-layer attention projection from pretrained ViT!
7:01 AM · Jun 6, 2026 · 2.9K Views
We are presenting FAE at #CVPR2026 morning 7:30-9:00 am today as well!
#134
We found a better latent space for generative models with single-layer attention projection from pretrained ViT!
Totally agreed! …and also maybe can check out FAE (https://arxiv.org/abs/2512.07829)
We are presenting FAE at #CVPR2026 morning 7:30-9:00 am today as well!
#134
We found a better latent space for generative models with single-layer attention projection from pretrained ViT!