How do you keep embeddings fresh for a constantly changing corpus?

Documents change daily, so stored embeddings go stale, but re-embedding everything is expensive. What's your refresh strategy?

25
0 comments

Comments (0)