Regarding persistent directory in chromadb vectorstore while using OpneAI Emeddings

Please, consider the following scenario:
We have some pdf files which are to be embedded via OpenAI embeddings in chromadb vectorstore. Initially, we define a persistent directory for storing the database on the system. Now we have more file to embed in the same directory. So if we call the function to create embeddings for the new file, would the already stored embedding be overwritten or would it append the new embeddings after the old ones.

Hi and welcome to the Developer Forum!

This is more a question for the ChromaDB team over on discord.gg/chromadb

But essentially, Chroma will append new data to your existing persistent storage unless you explicitly remove the old one.

1 Like