3 ms·

Once I've got embeddings my naïve next step would be to do cosine similarity for comparisons/search/anything that requires a distance. I see they do that in som

by IanCal 4y ago

Once I've got embeddings my naïve next step would be to do cosine similarity for comparisons/search/anything that requires a distance. I see they do that in some examples. Is that the standard approach these days? Are there newer default approaches that tend to work better?

visarga 4y ago
You can also apply clustering or train a classification model based on embeddings.
minimaxir 4y ago
Even better, the vectors come pre-unit normalized so you can just do the dot product for cosine similarity.