Navigating the Geometry of Food
You're looking at 250 food ingredients from cuisines around the world, mapped into 3D space by how they relate to each other. Ingredients that appear together in recipes cluster together. Ingredients that share flavor chemistry pull closer.
Three embedding models — COOC (co-occurrence in recipes), CORE (blended), and CHEM (flavor chemistry) — each organize the same ingredients differently. Drag the slider to watch the clusters reshape in real time.
Based on the paper by Radzikowski & Chen (2026), trained on 4.14M recipes across 7 languages.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。