it can be demonstrated that The easy pre-teaching job of predicting which caption goes with which image is surely an effective and scalable way to find out SOTA graphic representations from scratch with a dataset of https://k2spiceshop.com/product/liquid-k2-on-paper-online/