Show HN: World Flavor Atlas. Radzikowski and Chen's food embeddings in 3D

Category: other

Tags: 3d-visualization, food-embeddings, interactive-data

Score: 7.0/10 (Innovation: 7, Technical: 8, Documentation: 7, Utility: 6)

World Flavor Atlas is an interactive 3D visualization of food ingredient embeddings from the Epicure paper, mapping 250 ingredients by recipe co-occurrence and flavor chemistry. It offers real-time cluster morphing between three models, convex hulls, and fuzzy search, making complex embedding data explorable and visually engaging. This project uniquely bridges AI/ML research with an intuitive, browser-based tool for food science and culinary exploration.

Target audience: data scientists, researchers, food enthusiasts

Repository: https://tomacco.github.io/world-flavor-atlas/ · HTML · MIT

View on Hacker News