Configure cosine similarity

Create a collection that measures angular similarity between vectors.

Cosine similarity measures the angular relationship between vectors, ignoring magnitude. It is well suited for text embeddings where semantic meaning matters more than vector length.

The following example creates a collection named text_embeddings configured to use cosine similarity as its distance metric.

Python

from actian_vectorai import VectorAIClient, VectorParams, Distance

# Connect to VectorAI DB server
with VectorAIClient("localhost:6574") as client:
    # Create collection with cosine similarity
    client.collections.create(
        "text_embeddings",  # Collection name
        vectors_config=VectorParams(size=128, distance=Distance.Cosine)  # Cosine metric
    )

To confirm the collection was created successfully, call client.collections.get_info("text_embeddings") and check that the status field returns Ready.