Vector Search with SQL

Re:Vector Search with SQL by leobg on Wednesday January 12, 2022 @04:25PM (#29912887)

Looks great. How does it compare to hnswlib? How many vectors can it handle for subsecond queries? Is it approximate or precise nearest neighbor?

[ Parent | on HN ]

Re:Vector Search with SQL by neumll on Wednesday January 12, 2022 @04:37PM (#29913021)

Thank you. The indexes are built on top of either Faiss, Hnswlib or Annoy depending on the settings - https://neuml.github.io/txtai/embeddings/configuration/#back...

I've primarily focused on single query times, which have averaged around 5ms - 25ms depending on the size of the index. Queries can be batched so query times wouldn't increase linearly. You could batch quite a few and still get subsecond response times.

All three libraries are approximate nearest neighbors but I know at least for Faiss, it can be configured to effectively be a precise query.

[ Parent | on HN ]

Re:Vector Search with SQL by leobg on Wednesday January 12, 2022 @04:47PM (#29913155)

Thank you. I’ll check it out!

[ Parent | on HN ]