Thanks. Solr vector search was available when I got started on this Dense Vector Search :: Apache Solr Reference Guide, but I didn’t use it because it didn’t have dimensions compatible with the OpenAI model I was using at the time.
I don’t know if that has improved, but even if it has, I will probably stick with my current architecture which utilizes the Weaviate vector database. I mean, they’ve got this thing down to a science – low latency, super fast results, a text2vec transformer specifically for OpenAI models that increases the efficiency of the searches, hybrid ( BM25) search option, hybrid alpha slider to determine the mix of bm25 vs. semantic search, etc… I can go on and on singing their praises. I almost always get back the correct context documents with minimal effort on my part, leaving me to focus on adjusting my model prompts.
Why would I want to even think about building all of that myself?