Neural networks learn to approx functions y = f(x). Inference through a trained network is akin to looking up with query (X), values (in latent spa

Neural networks learn to approx functions y = f(x). Inference through a trained network is akin to looking up with query (X), values (in latent spa