picovdb 0.1.1

No project description provided

picovdb

---

An extremely fast, ultra-lightweight local vector database in Python.

"extremely fast": sub-millisecond query

"ultra-lighweight": One file with Numpy and one optional dependency faiss-cpu.

Install

pip install picovdb

Usage

Create a db:

(Use SentenceTransformer embedding as example)

from sentence_transformers import SentenceTransformer
from picovdb import PicoVectorDB

CHUNK_SIZE = 256
model = SentenceTransformer('all-MiniLM-L6-v2')
dim = model.get_sentence_embedding_dimension()

with open('A_Christmas_Carol.txt', encoding='UTF8') as f:
    content = f.read()
    num_chunks = len(content) // CHUNK_SIZE + 1
    chunks = [content[i * CHUNK_SIZE: (i + 1) * CHUNK_SIZE] for i in range(num_chunks)]
    embeddings = model.encode(chunks)
    data = [
        {
            "_vector_": embeddings[i],
            "_id_": i,
            "content": chunks[i],
        }
        for i in range(num_chunks)
    ]
    db = PicoVectorDB(embedding_dim=dim, storage_file='_acc')
    db.upsert(data)
    db.save()

Query

db = PicoVectorDB(embedding_dim=dim, storage_file='_acc')
txt = "Are there no prisons? Are there no workhouses?"
emb = model.encode(txt)
q = db.query(emb, top_k=3)
print('query results:', q)

Benchmark

Embedding Dim: 1024.

Hardware: M3 MacBook Air

1. Pure Python:

   • Inserting 100,000 vectors took about 0.5s
   • Doing 100 queries from 100,000 vectors took roughly 0.6s (0.006s per quiry).

2. With FAISS(cpu):

   • Inserting 100,000 vectors took 110s
   • Doing 100 queries from 100,000 vectors took 0.05s (0.0005s or 0.5 millisecond per quiry).
   • Doing 1000 queries from 100,000 vectors in batch mode took 0.2s (0.0002s or 0.2 millisecond per quiry).

Hardware: PC with CPU Core i7-12700k and old-gen M2 Nvme SSD

1. Pure Python:

   • Inserting 100,000 vectors took about 0.7s
   • Doing 100 queries from 100,000 vectors took roughly 1.3s (0.013s per quiry).

2. With FAISS(cpu):

   • Inserting 100,000 vectors took 50s
   • Doing 100 queries from 100,000 vectors took 0.05s (0.0005s or 0.5 millisecond per quiry).
   • Doing 1000 queries from 100,000 vectors in batch mode took 0.3s (0.0003s or 0.3 millisecond per quiry).