search

chroma-core / chroma

the AI-native open-source embedding database
https://www.trychroma.com/
Apache License 2.0
15.4k stars 1.29k forks source link

Help Needed - Query About ChromaDB #745

Closed anubhav562 closed 1 year ago

anubhav562 commented 1 year ago

Hey ChromaDB Team and Users,

I just came across chromaDB and wanted to know how do you store the vectors on the backend? Is it in-memory, is it completely disk based or partially disk and partially in-memory.

We have a use case and a document collection of over 2 billion documents and we are looking at different open source and private alternatives.

Can anyone guide me if chromaDB is at a stage where it can handle billions of documents at the moment or not ?

Please let me know!

Thanks, Anubhav

jeffchuber commented 1 year ago

@anubhav562 the current index type we support requires all vectors be in memory for graph traversal.

AvleenKour1 commented 1 year ago

Hi team , I have a usecase where in i am adding the documents into the collection and later querying on the collection multiple times .

i want to know how can i store the collection and next time ask the question to same collection in other notebook.

jeffchuber commented 1 year ago

@AvleenKour1 either use chroma in persistent mode or client-server mode. More info here: https://docs.trychroma.com/usage-guide

jeffchuber commented 1 year ago

Closing this as it is stale. Please let me know if anything else pops up here and we can re-open it.