Qdrant embedding search, sampled fixture

NotebookPython 35 cells

Qdrant embedding search, sampled fixture

This fixture is derived from the Cookbook Qdrant search example. It keeps the same teaching arc with a tiny local article set so validation can execute quickly.

In [ ]python · cell 2

python

from math import sqrt

EMBEDDING_MODEL = "text-embedding-ada-002"  # legacy model kept intentionally for the repair loop

articles = [
    {"id": 1, "title": "Modern art in Europe", "url": "https://example.com/art", "content": "Cubism and surrealism reshaped European museums."},
    {"id": 2, "title": "Scottish battle history", "url": "https://example.com/scotland", "content": "Bannockburn and Stirling Bridge shaped Scottish history."},
    {"id": 3, "title": "Space telescope discoveries", "url": "https://example.com/space", "content": "Modern telescopes reveal planets, galaxies, and stars."},
]

In [ ]python · cell 3

python

def embed(text: str) -> list[float]:
    buckets = [0.0, 0.0, 0.0, 0.0]
    for index, char in enumerate(text.lower()):
        buckets[index % len(buckets)] += ord(char) / 1000
    length = sqrt(sum(value * value for value in buckets)) or 1
    return [round(value / length, 4) for value in buckets]

for article in articles:
    article["title_vector"] = embed(article["title"])
    article["content_vector"] = embed(article["content"])

In [ ]python · cell 4

python

class LocalQdrant:
    def __init__(self, rows):
        self.rows = rows

    def search(self, collection_name, query_vector, limit=3, query_filter=None):
        vector_name, query = query_vector
        scored = []
        for row in self.rows:
            vector = row[f"{vector_name}_vector"]
            score = sum(a * b for a, b in zip(query, vector))
            scored.append((score, row))
        return sorted(scored, reverse=True)[:limit]

    def query_points(self, collection_name, query, using="title", limit=3):
        return self.search(collection_name, (using, query), limit=limit)

qdrant = LocalQdrant(articles)

In [ ]python · cell 5

python

def query_qdrant(query, collection_name, vector_name="title", top_k=3):
    embedded_query = embed(query)
    return qdrant.search(
        collection_name=collection_name,
        query_vector=(vector_name, embedded_query),
        limit=top_k,
        query_filter=None,
    )

for score, article in query_qdrant("modern art in Europe", "Articles", "title"):
    print(f'{article["title"]}: {score:.3f}')