98 pointsby todsacerdoti7 months ago4 comments

JKCalhoun7 months ago
I want to do this for some cut of Wikipedia. If there is already a RAG for Wikipedia, that would be cool. I want my own ... I don't know ... is it the Encyclopedia Galactica? (Or Hitchhiker's Guide to Earth and its Inhabitants?)
- PhilippGille7 months ago
  Here's a simple example using a part of a DBpedia dataset, but you can download other Wikipedia datasets and import those: https://github.com/philippgille/chromem-go/tree/v0.7.0/examp...
  (Disclosure: I'm the author)
  - JKCalhoun7 months ago
    Very cool — thanks. I'll play with it.
another_twist7 months ago
Neat. Does ollama have native support for storing documents ? I dont think the author wrote anything custom here (and I mean it in a good way)
- coolspot7 months ago
  The Open-WebUI has a built-in vector database and document storage.
danlitt7 months ago
Examples would be good - I'm not going to run all that locally just to check whether it's useful. (LLM helpers rarely are)
- hackacad7 months ago
  Yes, I will add some examples within the next days.
alexey-salmin7 months ago
I hope now I'll finally figure out how to patch KDE2 for FreeBSD
- VTimofeenko7 months ago
  In that case the RAG should be based on an anime dataset