2 ms·
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.
by rolisz 3y ago
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.
- moffkalast 3y agoIs 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 3y agoEven PrivateGPT does that, using Chroma as vector DB