Hacker News Reader

One thing I have been wondering about is how, concretely, in practice embeddings will be used. Will companies create embeddings for most / all content that they have? Will they create an embedding for every sentence, paragraph or page of text (or all of those). Will they store hierarchies of embeddings? Do they then store the original text so they can invert the process (that seems obvious)?

View on HN

User/Item/Query embeddings are the most common. That way you can generate per-user recommendations, or search results for a given query (with personalization using side information). Video will be interesting, once we have video embeddings (maybe this exists already). It depends on the use-case but a few of your ideas are certainly possible. Generally I've seen them at a coarse rather than fine level, but I'm sure that's out there too.

This looks like a good overview if you want to read about it: https://recsysml.substack.com/p/two-tower-models-for-retriev...

Reply on news.ycombinator.com