>>20082693Probably Lucinda too anon, if you know what I mean
>>20082689>The latent space is completely defined by the training set. The reverse diffusion process that walks you backwards from latent space noise to the bulk of the distribution where the original data liesWhen we use our eyes we train our neurons on the world around us. In our latent space, some of us can conjure a picture from words. I'm struggling to see how it is any different. A good prompter starts to get a feeling for the lay of the land in the latent space and can then shape the output like a sculptor carving a stone.
