Skip to content

Releases: FuxiVirtualHuman/MAE-Face

MAE-Face (ViT-Base) pre-trained models

25 Jan 03:37
Compare
Choose a tag to compare

the ViT-Base model pre-trained on face images
_pretrain: trained with per-patch normalization, which is suitable for representation and downstream tasks.
_visualize: trained without per-patch normalization, which is suitable for reconstruction and visualization.