Qdrant Vector Database

Use Qdrant as a vector database for your Knowledge Base.

Setup

Follow the instructions in the Qdrant Setup Guide to install Qdrant locally. Here is a guide to get API keys: Qdrant API Keys.

Example

1import os
2import typer
3from typing import Optional
4from rich.prompt import Prompt
5
6from kern.agent import Agent
7from kern.knowledge.knowledge import Knowledge
8from kern.vectordb.qdrant import Qdrant
9
10api_key = os.getenv("QDRANT_API_KEY")
11qdrant_url = os.getenv("QDRANT_URL")
12collection_name = "thai-recipe-index"
13
14vector_db = Qdrant(
15    collection=collection_name,
16    url=qdrant_url,
17    api_key=api_key,
18)
19
20knowledge_base = Knowledge(
21    vector_db=vector_db,
22)
23
24def qdrant_agent(user: str = "user"):
25    agent = Agent(
26        knowledge=knowledge_base,
27        debug_mode=True,
28    )
29
30    while True:
31        message = Prompt.ask(f"[bold] :sunglasses: {user} [/bold]")
32        if message in ("exit", "bye"):
33            break
34        agent.print_response(message)
35
36if __name__ == "__main__":
37    knowledge_base.insert(
38        url="https://kern-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
39    )
40
41    typer.run(qdrant_agent)

Async Support ⚡

Qdrant also supports asynchronous operations, enabling concurrency and leading to better performance.

1import asyncio
2
3from kern.agent import Agent
4from kern.knowledge.knowledge import Knowledge
5from kern.vectordb.qdrant import Qdrant
6
7COLLECTION_NAME = "thai-recipes"
8
9# Initialize Qdrant with local instance
10vector_db = Qdrant(
11    collection=COLLECTION_NAME,
12    url="http://localhost:6333"
13)
14
15# Create knowledge base
16knowledge_base = Knowledge(
17    vector_db=vector_db,
18)
19
20agent = Agent(knowledge=knowledge_base)
21
22if __name__ == "__main__":
23    # Load knowledge base asynchronously
24    asyncio.run(knowledge_base.ainsert(
25            url="https://kern-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
26        )
27    )
28
29    # Create and use the agent asynchronously
30    asyncio.run(agent.aprint_response("How to make Tom Kha Gai", markdown=True))

Tip

Using aload() and aprint_response() with asyncio provides non-blocking operations, making your application more responsive under load.

Qdrant Params

Name	Type	Default	Description
`collection`	`str`	-	Name of the Qdrant collection
`embedder`	`Embedder`	`OpenAIEmbedder()`	Embedder for embedding the document contents
`distance`	`Distance`	`Distance.cosine`	Distance metric for similarity search
`location`	`Optional[str]`	`None`	Location of the Qdrant database
`url`	`Optional[str]`	`None`	URL of the Qdrant server
`port`	`Optional[int]`	`6333`	Port number for the Qdrant server
`grpc_port`	`int`	`6334`	gRPC port number for the Qdrant server
`prefer_grpc`	`bool`	`False`	Whether to prefer gRPC over HTTP
`https`	`Optional[bool]`	`None`	Whether to use HTTPS
`api_key`	`Optional[str]`	`None`	API key for authentication
`prefix`	`Optional[str]`	`None`	Prefix for the Qdrant API
`timeout`	`Optional[float]`	`None`	Timeout for Qdrant operations
`host`	`Optional[str]`	`None`	Host address for the Qdrant server
`path`	`Optional[str]`	`None`	Path to the Qdrant database
`fastembed_kwargs`	`Optional[dict]`	`None`	Additional kwargs passed to `SparseTextEmbedding`.