Leehom Hu

@stingh711

All cardsCollections

From personal experience, I’d opt for 1536 dimensional embeddings since it’s gotten me way better results.

Postgres as a search engine — any blockers?

While semantic search is trendy, good old lexical search is still the backbone. Semantic techniques can improve results, but they work best when added to a solid text-based search foundation.

Postgres as a search engine — any blockers?

llm

It’s often beneficial to overlap your chunks to mitigate semantic gaps. My go-to settings are usually a chunk size of 256-304 tokens with 32 tokens of overlap. I’ve found this to work quite well, but your mileage may vary.

Building RAG with Postgres — any blockers?

rag, llm

The Chinese symbol for crisis 危机 is composed of two elements: one signifies danger and the other opportunity.

Random