AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Is Distributing All That Information Increases Precision?
Most of that information is redundant. But then what yannik was saying was interesting, does distributing all that information increase noise? I'm also incoding it like a different way over here or some uncertainty in the data. Some one training sample is a bit Different than others right? That might be quite far off. As as for the number of latent dimensions and so on, i mean, i agree with keith, i think having a big latent space and also having big weights may be more of a of a necessity for training than it really is for foreign coding.