Hacker News new | ask | show | jobs
by AvaRobinson22 1001 days ago
What large language models are you using? And database with API?
2 comments

Hi, great questions! Swirl currently uses the spaCy LLM for it similarity matching (the reranking of unified results lists from multiple sources).

If you mean connecting to various databases as content sources to search, yes, Swirl can do that (e.g. PostgreSQL as a source, etc.). Does that help?

Okay, but how does it handles indexing and query. Suppose we have a large database, what are the performance tradeoffs ? First you will do an SQL search to load data into Swirl ? Or is there something different?
We currently use Spacy Large English Core Model, we've experimented w/ others. We can store our meta data in many of the DBs supported by Django. We have a number of deployments on Postgres. We can also support Searching SQL DBs Using a flexible query template.
Okay, just asked a question in the same thread. About postgres.