Search papers, labs, and topics across Lattice.
1
0
2
9
Representation-Pivoted Autoencoders enable diffusion models to generate and edit images with higher fidelity by learning a compressed latent space that preserves the semantics of pre-trained visual representations.
