Can a vector database contain multiple collections

[UdayanSL]

new_db = Chroma.from_documents(
collection_name="test_collection",
documents=[original_doc],
embedding=OpenAIEmbeddings(), # using the same embeddings as before
ids=[document_id],
)
new_db = Chroma.from_documents(
collection_name="foo_collection",
documents=[second_doc],
embedding=OpenAIEmbeddings(), # using the same embeddings as before
ids=[new_document_id],
)

Can I define different collections like this? Will these collections be saved together under a common database (which is new_db in this case)?

OR

Will I have to define them separately into two different databases??

new_db = Chroma.from_documents(
collection_name="test_collection",
documents=[original_doc],
embedding=OpenAIEmbeddings(), # using the same embeddings as before
ids=[document_id],
)
new_db_foo = Chroma.from_documents(
collection_name="foo_collection",
documents=[second_doc],
embedding=OpenAIEmbeddings(), # using the same embeddings as before
ids=[new_document_id],
)

If the first case is true then how do I individually access the collections when required?? Please help urgently.

[williamjeong2]

I'm not familiar with chroma db, but it's possible for faiss. In memory, however, you'll need to combine one or more vector databases before using them.

I hope my answer helps.

[YLFxGen]

I used Qdrant but Chroma should be similar.

1. I used Qdrant here. Check Chroma doc to connect to db and create two collections.

client = QdrantClient(path=config.qdrant_dbpath)
client.recreate_collection(
    collection_name=config.qdrant_collection_name,
    vectors_config=VectorParams(size=1536, distance=Distance.DOT),
)

2. Connect to the collection you create and use add_documents instead of from_documents. I used Qdrant here, but I think chroma should be similar.

qdrant_store = Qdrant(
        client=QdrantClient(path=config.qdrant_dbpath),
        embeddings=config.openai_embedding_model,
        collection_name=config.qdrant_collection_name,
)
qdrant_store.add_documents(documents=chunk_docs)