Pinecone, Weaviate, Qdrant 비교

벡터 데이터베이스를 선택하는 이유

파인콘

위비에이트

큐드런트

Pinecone Code Example

from pinecone import Pinecone
pc = Pinecone(api_key='...')
index = pc.Index('docs')

index.upsert([
    ('id-1', vec1, {'source': 'a.pdf'}),
    ('id-2', vec2, {'source': 'b.pdf'})
])

results = index.query(
    vector=query_vec,
    top_k=5,
    filter={'source': 'a.pdf'}
)

Weaviate Code Example

import weaviate
client = weaviate.Client('http://localhost:8080')

client.batch.add_data_object(
    {'text': '...', 'source': 'a.pdf'},
    class_name='Doc',
    vector=vec1
)

results = (
    client.query.get('Doc', ['text', 'source'])
    .with_near_vector({'vector': query_vec})
    .with_limit(5)
    .do()
)

Qdrant Code Example

from qdrant_client import QdrantClient
from qdrant_client.models import PointStruct

client = QdrantClient('http://localhost:6333')

client.upsert(
    collection_name='docs',
    points=[PointStruct(id=1, vector=vec1, payload={'source': 'a.pdf'})]
)

results = client.search(
    collection_name='docs',
    query_vector=query_vec,
    limit=5
)

파인콘을 선택할 때

위비에이트를 선택할 때

큐드런트를 선택할 때

pgvector도 잊지 마십시오

CREATE EXTENSION vector;
CREATE TABLE docs (id SERIAL PRIMARY KEY, text TEXT, embedding vector(1536));
CREATE INDEX ON docs USING ivfflat (embedding vector_cosine_ops);

SELECT text FROM docs ORDER BY embedding <=> $1 LIMIT 5;

다른 선택지

마이그레이션의 고충

어느 것이 직접 호스팅을 우선합니까?

복습