Regarding use case for qdrant vector insest

General
1

Where can I find comprehensive documentation for the functions of vectordb (like qdrant) to know all functions and what they do?

For example: How is insert or upsert function of qdrant is going to process each Document I send? What all will be my fields in the point for metadata and so on

Also, why are there filters in insert and upsert function of qdrant, not used anyways?
def insert(self, documents: List[Document], filters: Optional[Dict[str, Any]] = None, batch_size: int = 10) → None:

In the same function I see code where:

document.embed(embedder=self.embedder)
payload={
                        "name": document.name,
                        "meta_data": document.meta_data,
                        "content": cleaned_content,
                        "usage": document.usage,
                    },

Why is the metadata of entire content of document embedded? What if some use cases only need the content to be embedded and use the meta_data to filter to perform retrieval? How can we do advanced search like this POST /collections/online_store/points/search as shown on qdrant official docs.
Also, give us the option for setting the doc_id and put default to doc_id = md5(cleaned_content.encode()).hexdigest(), since I’ve a uuid which I can use as a doc_id and my upsert logic depends on it to keep it less redudant
If you can incoporate these things, it’ll be a great wrapper for qdrant. Currently, I feel qdrant API doc is pretty intuitive and easier to follow

Thank you for your time.

2

Also, when trying this function:

def generate_with_vector_search(vectordb: Qdrant) -> str:
    knowledge_base = AgentKnowledge(
        vector_db=vectordb,
        num_documents=10
    )
    knowledge_base.load(recreate=False)

I’m getting error that for document_list in self.document_lists: NotImplementedError

referenced at venv\Lib\site-packages\phi\knowledge\agent.py",

What alternative do I have instead of this, will this be fixed in 3.0.0?

3

Hi @aatish
Thank you for reaching out and using Phidata! I’ve tagged the relevant engineers to assist you with your query. We aim to respond within 48 hours.
If this is urgent, please feel free to let us know, and we’ll do our best to prioritize it.
Thanks for your patience!

4

@aatish you can take a look at all the functions of qdrant from here phidata/phi/vectordb/qdrant/qdrant.py at main · phidatahq/phidata · GitHub

We definitely need to improve the docs on this front and include it as for filters not being used that is a good catch. I will bring this up to the team and try to have a fix soon

5

Thank you for your response @manthanguptaa!
Can you shed some light on this as well? Are you working on this logic or will it be the same in the future releases?

6

Hey @aatish, we don’t embed the meta-data with the content. We only insert it in the table column. You can pass the meta-data to the filter column to use it for retrieval

7

Hi @manthanguptaa
def search(self, query: str, limit: int = 5, filters: Optional[Dict[str, Any]] = None) -> List[Document]:

THis is the function of search I found, and when going through the code, I saw that FILTERS ARE NOT BEING USED at all for performing search, maybe I’m wrong but you should look into your code, also hard codingdoc_id = md5(cleaned_content.encode()).hexdigest() is not the way I feel.

I’m not using Phi anymore because of the poor integration, misleading documents and lack of support. I just wanted to get this clarified.

8

Hey @aatish!

I truly appreciate you taking the time to explore our framework, and I’m really sorry if your experience wasn’t the best. We’ve been working hard to improve things, and as part of that journey, we’ve recently rebranded to Agno! Along with the new name, we’ve made a lot of enhancements to the framework. We are now working on improving our knowledge bases and vector database integrations.
I’d love for you to give it another shot and see how much we’ve improved. Your thoughts mean a lot to us, and we’re always here to listen and make things better. Let me know if there’s anything I can do to help!