Skip to content
Better HN
Top
Best
Ask
Show
New
Jobs
Search
⌘K
0 points
rolisz
3y ago
0 comments
Save
Share
Well you wouldn't input the whole Vault to the model, you would use embeddings to find the content that is most relevant to the question being asked.
0 comments
2 comments · 1 top-level
top
newest
oldest
moffkalast
3y ago
· 1 in thread
Is that actually a thing yet? Proper vector DB integration? I sure would like to see some demos of that, as it's been hyped up a lot but I haven't really seen anyone deploy anything proper with it yet.
rolisz
OP
3y ago
Even PrivateGPT does that, using Chroma as vector DB
j
/
k
navigate · click thread line to collapse