Find answers from the community

Updated 3 months ago

hello everyone, I'm working with qdrant

hello everyone, I'm working with qdrant vector store but it looks like the embedding of new nodes is not saved in qdrant but the retrieve works anyway...can someone explain a bit more how qdrant works? I followed the example in the llama index documentation but the embedding vector of documents can not be seen anyware
W
m
8 comments
Hey!
So qdrant does not show embeddings on the UI as embeddings are very large matrix.

But if you scroll a bit down on the UI, it shows Copy Vectors or something related to that at the end of node. From there you can copy the vector for that particular node
thanks for reply! I tryied to copy the embedding but it looks nothing inside...also if a call

POST collections/portal/points/scroll
{
"limit": 10
}

from console, the vector field is empty
This is on the UI right?
With this endpoint this becomes more of Qdrant question and less LlamaIndex πŸ˜…

not much idea for Qdrant API
i think is more a problem with llama index qdrant works pretty well the only things is that when I create a vectore index in llama index nodes in qdrant are populated but the vector field is empty
client = QdrantClient(
url=cloud_qdrant,
api_key=api_key,
)
vector_store = QdrantVectorStore(collection_name=self.engine.url.database, client=self.client)
storage_context = StorageContext.from_defaults(vector_store=vector_store)
vect_index = VectorStoreIndex(nodes=nodes, storage_context=storage_context)
this is my code
nodes are llama index TextNodes
finally it was a wrong command in qdrant console, the "with_vectors" tag was missing

POST collections/portal/points/scroll
{
"limit": 10,
"with_vectors": true
}

now I can see the vectors πŸ˜… thanks again
Add a reply
Sign up and join the conversation on Discord