Log in
Log into community
Find answers from the community
View all posts
Related posts
Did this answer your question?
π
π
π
Powered by
Hall
Inactive
Updated 4 months ago
0
Follow
Hi guys, I'm getting mad... Is there a
Hi guys, I'm getting mad... Is there a
Inactive
0
Follow
P
Ponce
4 months ago
Β·
Hi guys, I'm getting mad... Is there a way to update a created vector store index adding documents with embeddings and all?
L
P
6 comments
Share
Open in Discord
L
Logan M
4 months ago
index.insert(document) ?
L
Logan M
4 months ago
Or do you mean updating existing documents
L
Logan M
4 months ago
You kind of need to have kept track beforehand while inserting to updating existing. You have to insert a document witht the same document ids as the original, usually with an ingestion pipeline
https://docs.llamaindex.ai/en/stable/module_guides/loading/ingestion_pipeline/#document-management
P
Ponce
4 months ago
I now have this:
chroma_collection = db.get_collection("quickstart")
vector_store = ChromaVectorStore(chroma_collection=chroma_collection)
storage_context = StorageContext.from_defaults(vector_store=vector_store)
vector_store_index = VectorStoreIndex.from_vector_store(
vector_store,
embed_model=embed_model,
)
print("Index loaded from storage.")
new_documents = load_new_documents(folder_path)
if new_documents:
# Parse documents into nodes
print("Parsing new documents into nodes...")
parser = SimpleNodeParser()
new_nodes = parser.get_nodes_from_documents(new_documents, show_progress=True)
# Add nodes to the existing index
print("Adding new nodes to the existing index...")
vector_store_index.insert_nodes(new_nodes)
print("New documents added to the index.")
P
Ponce
4 months ago
the fact is that in this way the embedder model never gets called, the nodes are added as "plain text" not as embeddings
L
Logan M
4 months ago
How can the emded model not get called?
insert_nodes() has to call the embed model π
Add a reply
Sign up and join the conversation on Discord
Join on Discord