Conceptual
Login

Patchifying a Latent Tensor into Transformer Tokens

the exact join between the convolutional VAE and the transformer denoiser — patch size is the compute knob everything else scales off

This Concept is waiting for its first lesson!

the exact join between the convolutional VAE and the transformer denoiser — patch size is the compute knob everything else scales off

Are you a teacher? Sign in to start contributing.

Sign In