Smudges: a retrospective

Doodles, captions for those doodles generated by Vision Transformer (ViT) deep learning models, and generated captions extended by autoregressive transformer language models, but in the form of a coffee table book, and inspired by Werner Herzog.