GLOSSARY TERM
What is Latent Space?
A compressed, high-dimensional representation of compressed data points.
Latent space maps structurally similar data points into close proximity. Traversing or interpolating within this multi-dimensional manifold allows generative models to synthesize novel, coherent data that shares the training set's core properties.
Manipulate Data Manifolds
Explore the depths of representation learning directly on the M1 stack.