Find answers from the community

Updated 3 months ago

Hello, What are the reasons when a query

Hello, What are the reasons when a query engine gives a "Empty response" when i tried to query? i am using MongoDB as a vector store and i use load_index_from_storage to retrieve the index.

13 comments

LLogan M

load index from storage isn't really for remote vector dbs

Probably you should be using VectorStoreIndex.from_vector_store(vector_store) to "load" if using mongodb atlas

HHardrion

i use mongodb locally, when i try to run query engine for the first time it works just like in this image

Attachment

HHardrion

but when i got rid of the VectorStoreIndex or just retrieve the data from the storage, i receive Empty response

HHardrion

Attachment

LLogan M

oh so you are only storing the index store in mongodb. There is still a vector_store and docstore objects that are being lost here because they aren't being persisted

HHardrion

i thought data will automatically persist in the DB as i have document with type vector_store being created

HHardrion

do you have advice?

LLogan M

I think you are misunderstanding how the storage works in llama-index

I would just pick a proper vector db integration (qdrant, chroma, weaviate, etc.) and use that

HHardrion

i use mongodb because i can use both docstore and vector store, and i seem to realize the problem now because in the db their are no embeddings being saved

LLogan M

yea, only mongodb atlas can save embeddings

LLogan M

but thats cloud only, not local

HHardrion

ahhhhh ok, thank you for the information

HHardrion

truly, because i would have spent many days trying to figure out why it was not saved

Add a reply