$XNY MM-Food-100K on Hugging Face A public 100k-sample multimodal food dataset from Codatta is live for research; an accompanying arXiv preprint details collection, schema, and use for food-AI tasks. Good showcase of the “contributor → open science” loop.