Configure dot product

Dot product measures both alignment and magnitude. For pre-normalized embeddings, it produces the same results as cosine similarity with better computational performance. Check your embedding model’s documentation to confirm whether its output vectors are normalized before choosing this metric. The following example creates a collection named normalized_embeddings configured to use dot product as its distance metric.

from actian_vectorai import VectorAIClient, VectorParams, Distance

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

import { VectorAIClient } from '@actian/vectorai-client';

async function main() {
    // Connect to VectorAI DB server
    const client = new VectorAIClient('localhost:6574');

    // Create collection with dot product
    await client.collections.create('normalized_embeddings', {
        dimension: 128,
        distanceMetric: 'DOT'
    });
}

main().catch(console.error);

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

Configure Euclidean distance Indexing overview

⌘I

Collections

Points

Vectors

Payload

Search

Filtering

Semantic search

Hybrid search

Distance metrics

Indexing

Configure dot product