Understanding Masked Autoencoders via Hierarchical Latent Variable Models

Open in new window