Connor Shorten - Research Scientist, Weaviate - ChatGPT, LLMs, Form vs Meaning
Vector Podcast
00:00
How to Catch an Elaskin Pollock With Vector Search
With vector search you may try to find what would what could be the most similar object but it may also be wrong right or uh in the case when the distance is so big that it's doesn't make sense anymore to consider this as a candidate. The idea of like how do you continually update your embedding models is fascinating and i haven't figured out how to make this work yet. If we were to insert new neighbors into the existing graph rightUh would that change anything? We are still in the realm of this hybrid search topic in a way, he says.
Transcript
Play full episode
